501 D2C growth benchmarks

A modelled negotiated micro-influencer post costs Rs 14,400 including fee, management and negotiation time, against Rs 512 for an ambassador post.

Modelled on stated assumptions: Rs 12,000 fee, 2 hours management, 1 hour negotiation at Rs 800, against commission plus 0.1 hours. · Ambassador programs: the middle ground between affiliate n influencer · Influencer and creator operations

Per acquired order that is Rs 2,880 against Rs 320, so the ambassador route is 9 times cheaper per order rather than 28 times cheaper per post.

Arithmetic on the same assumptions: 5.0 orders per negotiated post against 1.6 per ambassador post. · Ambassador programs: the middle ground between affiliate n influencer · Influencer and creator operations

Of 500 self-serve signups a modelled 150 post once and 50 post repeatedly, leaving 350 who never post and cost nothing in commission.

Modelled on stated assumptions: 30% single activation, 10% repeat activation. · Ambassador programs: the middle ground between affiliate n influencer · Influencer and creator operations

Every SB&R referral link belongs to someone who has already bought, which is the 1 qualifying condition an ambassador program needs and the reason self-serve signup works.

SB&R referral link issuance rule. · Ambassador programs: the middle ground between affiliate n influencer · Influencer and creator operations

A PayPal and Comscore study found 27% of shoppers would abandon a cart to go and search for a voucher code.

PayPal and Comscore, cited by Zuko · Why automatic discounts outperform code-entry every time · Urgency, scarcity and flash

Baymard puts average documented cart abandonment at 70.22% across 50 studies, so checkout traffic is the scarcest thing a store has.

Baymard Institute cart abandonment research, cited by Growth Engines · Why automatic discounts outperform code-entry every time · Urgency, scarcity and flash

Per 1,000 checkout sessions a visible code field costs a modelled Rs 44,906 in lost contribution, unintended discounts and affiliate commission.

Modelled on stated assumptions: 27% leave to search, 40% never return, 30% baseline conversion, Rs 1,800 average order value. · Why automatic discounts outperform code-entry every time · Urgency, scarcity and flash

That is 1.66 times the Rs 27,000 it would cost to apply a 5% automatic discount to every one of those orders instead.

Arithmetic on the same modelled assumptions. · Why automatic discounts outperform code-entry every time · Urgency, scarcity and flash

A reason-led flow recovering 8% of 700 abandoned carts produces 56 orders and Rs 35,280 of contribution with no discount cost at all.

Modelled on stated assumptions: 1,000 checkout sessions, 70% abandonment, Rs 630 contribution per order. · Cart abandonment flows that don't beg · D2C growth experiments

An escalating discount flow recovers 13% instead, but at an average 15% discount the same 91 orders contribute only Rs 32,760 before any other effect.

Modelled on the same assumptions at Rs 1,800 average order value. · Cart abandonment flows that don't beg · D2C growth experiments

If 8% of the 300 shoppers who would have completed learn to abandon deliberately, that costs a further Rs 8,640, taking the discount flow to Rs 24,120 net.

Modelled on 24 strategic abandoners receiving a 20% discount they would not otherwise have had. · Cart abandonment flows that don't beg · D2C growth experiments

The reason-led flow therefore wins by Rs 11,160, or 46%, despite recovering 5 percentage points fewer carts.

Arithmetic on the same modelled assumptions. · Cart abandonment flows that don't beg · D2C growth experiments

In a modelled 320-pincode order map, the top decile of postcodes carries 49% of revenue while the bottom decile carries 2.9%, a ratio of 17 to 1.

Modelled on a Zipf distribution, exponent 0.8, across 320 pincodes and 900 monthly orders at a flat average order value. · Why your best customers are in six postcodes and you don't know which · Hyperlocal and geo marketing

The same model puts 24.8% of revenue in the top 6 postcodes, which is 1.9% of the postcode list.

Modelled on a Zipf distribution, exponent 0.8, across 320 pincodes. · Why your best customers are in six postcodes and you don't know which · Hyperlocal and geo marketing

Shopify's prebuilt regional report, Total Sales by Billing Location, groups by billing country and region, so 0 of the default views show shipping postcode.

Shopify Help Center, sales reports · Why your best customers are in six postcodes and you don't know which · Hyperlocal and geo marketing

86 of 320 postcodes in the modelled map produce under 1 order a month, so most of the list can never support a conclusion about anything.

Modelled on a Zipf distribution, exponent 0.8, across 320 pincodes and 900 monthly orders. · Why your best customers are in six postcodes and you don't know which · Hyperlocal and geo marketing

Barter of a Rs 1,800 product at 40% gross margin costs Rs 1,135 including shipping, or 63% of the Rs 1,800 value the creator perceives.

Modelled on stated assumptions: Rs 1,800 list price, Rs 55 shipping, cost of goods derived from gross margin. · Barter, fee, or commission: a decision table · Influencer and creator operations

At 70% gross margin the same barter costs 33% of perceived value, so barter efficiency is entirely a function of margin.

Arithmetic on the same assumptions. · Barter, fee, or commission: a decision table · Influencer and creator operations

At 30% gross margin barter saves only Rs 485 against perceived value while adding fulfilment, shipping and the risk of no post.

Arithmetic on the same assumptions. · Barter, fee, or commission: a decision table · Influencer and creator operations

Commission-only cannot match a flat fee at typical creator volumes, since a Rs 8,000 rate against Rs 6,480 of attributed revenue implies a 123% commission.

Modelled on stated assumptions: 3.6 attributed orders at Rs 1,800 average order value. · Barter, fee, or commission: a decision table · Influencer and creator operations

At a 15% chance of detection per exposure, 55.6% of shoppers have caught a resettable timer after 5 exposures and 80.3% after 10.

Modelled on a per-exposure detection probability of 0.15, compounded as one minus 0.85 to the power of exposures. · Countdown timers are dead. This replaced them. · Urgency, scarcity and flash

Detection is one-way, so timer credibility decays monotonically with exposure and reaches 96.1% detected by the 20th exposure.

Modelled on the same per-exposure detection probability. · Countdown timers are dead. This replaced them. · Urgency, scarcity and flash

A client-side timer stores its deadline in browser storage, so a private window or a cleared cache restores the full duration in 1 action.

Standard client-side countdown implementation behaviour. · Countdown timers are dead. This replaced them. · Urgency, scarcity and flash

A server-held window returns the same remaining time to all 3 common detection routes, refresh, private window and second device, so the test confirms it.

Derived from server-authoritative deadline behaviour, where 1 stored value serves every request. · Countdown timers are dead. This replaced them. · Urgency, scarcity and flash

A creator who can see 90 link clicks and is told they drove 3.6 orders is looking at a 4% conversion rate, which reads as underreporting without the path shown.

Modelled on stated assumptions: 3% click-through on a 3,000 reach and 4% conversion. · The creator payout report that ends the monthly argument · Influencer and creator operations

On a 200-creator roster a 12% monthly dispute rate consumes 24 disputes and 18 hours, which is 11% of a manager's month or Rs 14,400.

Modelled on stated assumptions: 0.75 hours per dispute at Rs 800 an hour. · The creator payout report that ends the monthly argument · Influencer and creator operations

Cutting that rate to 4% saves 12 hours and Rs 9,600 a month on the same roster.

Arithmetic on the same assumptions. · The creator payout report that ends the monthly argument · Influencer and creator operations

Korant resolves each sale through 5 checks in a fixed priority order, and naming the check that fired is what makes a payout line auditable.

Korant sale attribution priority order. · The creator payout report that ends the monthly argument · Influencer and creator operations

A minimum live period is missing from most creator agreements, so a post taken down after 48 hours has breached nothing at all.

Derived from standard creator agreement structures, which specify publication but not duration. · Creator contracts in plain language · Influencer and creator operations

An attribution window of 7 days against 30 produces materially different payouts on identical work, so the number has to be fixed before the campaign.

Derived from cookie-based attribution windows. · Creator contracts in plain language · Influencer and creator operations

Leaving approval rounds unspecified means unlimited, against a modelled 14.4 calendar days already consumed by a script-style brief at 3 days a round.

Modelled on 4.8 expected revision triggers at 3 calendar days per round. · Creator contracts in plain language · Influencer and creator operations

SB&R coins redeem as a capped checkout discount and are never paid out as cash, so 1 of the 7 clauses, payment, cannot be satisfied by offering coins.

SB&R coin redemption rule. · Creator contracts in plain language · Influencer and creator operations

A modelled Rs 4,000 build for per-pincode delivery estimates breaks even at 6.3 incremental orders, once, at Rs 630 contribution per order.

Modelled on stated assumptions: 5 hours at Rs 800 per hour, Rs 1,800 average order value, 35% contribution. · Delivery time is a marketing message you are not sending · Hyperlocal and geo marketing

Spread over a year at 900 orders a month, that build needs a 0.06% sustained conversion lift to pay for itself.

Arithmetic on the same assumptions: 6.3 orders against 10,800 annual orders. · Delivery time is a marketing message you are not sending · Hyperlocal and geo marketing

A 10% sitewide discount costs Rs 2,430 per incremental order and needs an 8% lift merely to produce them, against a one-off Rs 4,000 for the estimate.

Modelled on stated assumptions: Rs 1,800 average order value, 8% incremental lift, 900 monthly orders. · Delivery time is a marketing message you are not sending · Hyperlocal and geo marketing

Because 49% of orders sit in the fastest decile of postcodes, an honest estimate only loses money if the slow-zone conversion penalty exceeds 2.16 times the fast-zone gain.

Modelled on a Zipf distribution, exponent 0.8, with the fastest decile at 49% of orders and the slowest six deciles at 22.7%. · Delivery time is a marketing message you are not sending · Hyperlocal and geo marketing

At a monthly discount cadence, a published calendar leaves a modelled full-price share of 48.7% while unforecastable timing leaves 97.4%.

Modelled on an exponential willingness-to-wait distribution with a 3 week patience constant, scaled by the share of shoppers who can forecast the next date. · Building a discount calendar your competitors cannot read · Urgency, scarcity and flash

Running the same event in the same fortnight each year keeps roughly 80% of shoppers able to forecast it, costing 10.2 points of full-price share against variable seasonal dates.

Modelled on the same distribution at an assumed 80% forecastable share against 30%. · Building a discount calendar your competitors cannot read · Urgency, scarcity and flash

The discount cadence is identical across all 5 predictability levels modelled, so the entire 48.8 point swing comes from forecastability alone.

Arithmetic on the same model holding the monthly cadence constant. · Building a discount calendar your competitors cannot read · Urgency, scarcity and flash

A fixed monthly cadence left unpublished still leaves 60% of shoppers able to forecast it, because the pattern is legible from behaviour without any announcement.

Modelled assumption on pattern inference from a regular unannounced cadence. · Building a discount calendar your competitors cannot read · Urgency, scarcity and flash

Modelled margin per attempted order falls from Rs 534 in a metro core zone to Rs 357 in a remote zone, leaving the remote zone at 67% of core profitability.

Modelled on stated assumptions: Rs 1,800 average order value, 35% gross contribution, zone shipping of Rs 55 to Rs 110, RTO of 6% to 22%. · Delivery zones as a growth channel, not a logistics table · Hyperlocal and geo marketing

A 22% return-to-origin rate costs roughly Rs 48 per attempted order in wasted forward and return freight alone, before any handling or restocking cost.

Modelled on stated assumptions: Rs 110 forward and Rs 110 return freight in a remote zone. · Delivery zones as a growth channel, not a logistics table · Hyperlocal and geo marketing

Shopify's prebuilt regional sales report groups by billing country and region, so 0 default views combine revenue with the shipping zone that produced it.

Shopify Help Center, sales reports · Delivery zones as a growth channel, not a logistics table · Hyperlocal and geo marketing

Raising the free-shipping threshold by Rs 200 in remote zones only recovers the margin gap if it lifts average order value by more than 11%.

Modelled on stated assumptions: Rs 177 margin gap per order between core and remote zones at a 35% contribution rate. · Delivery zones as a growth channel, not a logistics table · Hyperlocal and geo marketing

Shopify sorts discounts into 3 classes and applies them in a fixed order: product first, then order on the reduced subtotal, then shipping last.

Shopify Help Center, combining discounts · The discount stacking rules that quietly kill your margin · Urgency, scarcity and flash

A 15% product discount, a 25% order discount and a 10% wallet redemption compound to a 46.6% giveaway on a Rs 2,000 order, against a 31.05% break-even.

Modelled on stated assumptions: Rs 2,000 list, 35% gross margin, Rs 79 shipping absorbed. · The discount stacking rules that quietly kill your margin · Urgency, scarcity and flash

That stack turns a Rs 700 gross margin order into a Rs 231.50 loss once free shipping is absorbed.

Modelled on the same assumptions, with COGS of Rs 1,300. · The discount stacking rules that quietly kill your margin · Urgency, scarcity and flash

Capping a wallet redemption on list price rather than on the running subtotal deepens the same stack from 46.6% to 50.2% and the loss from Rs 231.50 to Rs 304.

Modelled on the same assumptions, varying only the base the wallet cap is computed against. · The discount stacking rules that quietly kill your margin · Urgency, scarcity and flash

A modelled 200,000-follower creator at 3% overlap reaches 6,000 relevant people, against 7,000 for a 20,000-follower creator at 35% overlap.

Arithmetic on stated assumptions: effective audience equals followers multiplied by overlap share. · Finding creators whose audience overlaps yours · Influencer and creator operations

Across a plausible overlap range of 3% to 35%, effective audience swings by a factor of 11.7 at constant follower count.

Arithmetic on the same assumptions. · Finding creators whose audience overlaps yours · Influencer and creator operations

On modelled Indian rate cards the same three creators produce a CAC of Rs 15,000, Rs 2,381 and Rs 1,042, in descending order of follower count.

Modelled on stated assumptions: Rs 90,000, Rs 12,000 and Rs 6,000 fees against 6.0, 5.0 and 5.8 attributed orders. · Finding creators whose audience overlaps yours · Influencer and creator operations

Geographic overlap is the one dimension a brand can measure from its own data, since every order already carries 1 shipping postcode.

Derived from Shopify order shipping address fields. · Finding creators whose audience overlaps yours · Influencer and creator operations

A 5 minute window reaches 35.2% of interested shoppers where the median decision time is 8 minutes, and 0.08% where the median is 3 days.

Modelled on an exponential decision-time distribution, share reached equals one minus two to the power of negative window over median. · The five-minute discount window: does it convert or annoy · Urgency, scarcity and flash

Reaching 80% of interested shoppers requires a window of 2.32 times the category's median decision time, whatever that median is.

Derived from the same exponential model: log base 2 of 5 equals 2.32. · The five-minute discount window: does it convert or annoy · Urgency, scarcity and flash

A window equal to the median decision time reaches exactly 50% of interested shoppers, by definition of the median.

Derived from the definition of a median decision time. · The five-minute discount window: does it convert or annoy · Urgency, scarcity and flash

A 24 hour window reaches 20.6% of shoppers in a 3 day median category, so even a full day is short for a considered purchase.

Modelled on the same exponential decision-time distribution. · The five-minute discount window: does it convert or annoy · Urgency, scarcity and flash

In a modelled 45-pincode demand map for one slow SKU, the top 3 pincodes carry 31.6% of that SKU's demand while representing 0.9% of the brand's 320-pincode map.

Modelled on a Zipf distribution, exponent 0.8, across the SKU's own 45-pincode order map. · Flash mechanics for slow-moving inventory · Urgency, scarcity and flash

That is 33.7 times more of the SKU's demand reached per unit of brand exposure to the discounted price.

Arithmetic on the same modelled distribution. · Flash mechanics for slow-moving inventory · Urgency, scarcity and flash

Clearing 400 units in 6 months at 900 monthly orders requires a 7.4% attach rate across every order the store takes.

Arithmetic on stated assumptions: 400 units, 5,400 orders over six months. · Flash mechanics for slow-moving inventory · Urgency, scarcity and flash

A rotation reaching 2.06% of orders would need a 360% attach rate to clear the same 400 units, so rotation cannot clear inventory on volume.

Arithmetic on the same assumptions against a modelled 20-week rotation cycle. · Flash mechanics for slow-moving inventory · Urgency, scarcity and flash

Fortnightly sitewide discounting moves 71.7% of demand off full price on a modelled 3 week patience constant, against 11.5% for quarterly sales.

Modelled on an exponential willingness-to-wait distribution with a 3 week patience constant, expected wait of half the discount period. · Flash sales without the brand damage · Urgency, scarcity and flash

Rotating one weekly window across 20 zones gives the brand 52 promotional events a year while any individual shopper's zone goes live 2.6 times.

Arithmetic on 52 weekly windows divided across 20 rotation zones. · Flash sales without the brand damage · Urgency, scarcity and flash

That rotation drops the modelled deferral share from 71.7% to 3.6% at an identical brand-level discount frequency of 52 events a year.

Modelled on the same 3 week patience constant, comparing a 2 week shopper-facing period against a 20 week one. · Flash sales without the brand damage · Urgency, scarcity and flash

Moving from monthly to quarterly sales cuts modelled deferral from 51.3% to 11.5%, which is the single largest lever available without reducing discount depth.

Modelled on the same exponential willingness-to-wait distribution. · Flash sales without the brand damage · Urgency, scarcity and flash

Shopify Functions must execute within a maximum of 11 million instructions, and exceeding it fails the run rather than degrading it.

Shopify Functions limitations and considerations · Running a flash window during a traffic spike · Urgency, scarcity and flash

Shopify's own guidance is to read from a metafield rather than make an external network call, replacing 1 network dependency with storage Shopify already operates.

Shopify, about network access for Shopify Functions · Running a flash window during a traffic spike · Urgency, scarcity and flash

A 10 minute spike of 20,000 sessions at 6 page views each is 200 requests a second, or 120,000 lookups for a per-request discount architecture.

Arithmetic on stated assumptions: 20,000 sessions over 10 minutes, 6 page views per session. · Running a flash window during a traffic spike · Urgency, scarcity and flash

The pre-computed pattern replaces those 120,000 reads with 1 metafield write per rotation, executed before any shopper arrives.

Derived from the pre-computed offer architecture described. · Running a flash window during a traffic spike · Urgency, scarcity and flash

Of 8 signals a Shopify store could gate a geo offer on, only 1 cannot be altered by the shopper without also changing where the order ships.

Trust boundary audit of Shopify cart, checkout and customer fields. · Geo-targeted offers that don't leak to the wrong customer · Hyperlocal and geo marketing

A discount code carries 0 geographic properties, so a code shared outside its intended zone applies at full value with no further effort.

Shopify discount code behaviour. · Geo-targeted offers that don't leak to the wrong customer · Hyperlocal and geo marketing

Cart attributes are writable through the public Ajax Cart API, putting exactly 1 HTTP request between a shopper and a geo-gated discount.

Shopify Ajax Cart API, cart update endpoint. · Geo-targeted offers that don't leak to the wrong customer · Hyperlocal and geo marketing

Gating on cart.deliveryGroups[0].deliveryAddress.zip means the only bypass is shipping the order into the target zone, which converts 100% of successful bypass attempts into genuine in-zone orders.

Shopify Discount Function input fields. · Geo-targeted offers that don't leak to the wrong customer · Hyperlocal and geo marketing

A modelled geo-density loop at a 1.10 coefficient and a 30 day cycle doubles in 218 days, against 373 for a content loop at 1.25 and 120 days.

Arithmetic on compound growth: doubling time equals cycle length times ln 2 divided by ln of the loop coefficient. · The growth loops that don't need paid media · D2C growth experiments

A referral loop at 1.15 with a 45 day cycle doubles in 223 days, almost identical to the weaker but faster geo-density loop.

Arithmetic on the same formula. · The growth loops that don't need paid media · D2C growth experiments

Raising the coefficient from 1.10 to 1.35 while the cycle stretches from 30 to 180 days makes doubling time worse, moving from 218 days to 416.

Arithmetic on the same formula across the two parameter pairs. · The growth loops that don't need paid media · D2C growth experiments

Cycle time is the only one of the 2 loop parameters a brand can usually shorten deliberately, since the coefficient is mostly a property of the product.

Derived from the components of each loop: coefficient depends on willingness to share, cycle on operational timing. · The growth loops that don't need paid media · D2C growth experiments

A brand running 8 sale days a quarter has 8.8% of days on sale, so a healthy cohort puts about 8.8% of orders in those windows and a trained one puts 45%.

Arithmetic on stated assumptions: 8 sale days in a 91 day quarter. · How often can you discount before shoppers wait for it · Urgency, scarcity and flash

That trained pattern is a 5.1x concentration ratio of orders per sale day against orders per ordinary day.

Arithmetic on the same assumptions. · How often can you discount before shoppers wait for it · Urgency, scarcity and flash

A cohort with a natural 60 day repurchase interval drifts to 91 days as it learns a quarterly cadence, cutting order frequency by 34.1%.

Modelled on stated assumptions: 60 day natural interval, 91 day sale period, linear blend by trained share. · How often can you discount before shoppers wait for it · Urgency, scarcity and flash

Combined with a 25% discount on those orders, revenue per customer per year falls 50.5%, from Rs 10,950 to Rs 5,415.

Arithmetic on the same assumptions at Rs 1,800 average order value. · How often can you discount before shoppers wait for it · Urgency, scarcity and flash

A single top postcode holds only about 9% of a modelled brand's weekly orders, so one live window can move roughly 18 baseline orders a week and no more.

Modelled on a Zipf distribution, exponent 0.8, across 320 pincodes and 900 monthly orders. · Hyperlocal flash sales without a physical store · Hyperlocal and geo marketing

At a 40% conversion lift inside the live zone, a 7 day window costs Rs 630 per incremental order against Rs 2,430 for a sitewide discount at 8% lift.

Modelled on stated assumptions: Rs 1,800 average order value, 10% discount depth, weekly baseline of 18 orders in the live zone. · Hyperlocal flash sales without a physical store · Hyperlocal and geo marketing

Rotating across the top 6 postcodes covers 25% of weekly order volume, which is the realistic ceiling on what a single rotation programme touches.

Modelled on a Zipf distribution, exponent 0.8, across 320 pincodes. · Hyperlocal flash sales without a physical store · Hyperlocal and geo marketing

The discount gates on cart.deliveryGroups[0].deliveryAddress.zip, which is 1 of Shopify's own delivery fields rather than a shopper-writable cart attribute.

FlashPin Shopify Discount Function implementation. · Hyperlocal flash sales without a physical store · Hyperlocal and geo marketing

Of a modelled 100 people who see a creator post, 40 tap the link and 25 search the brand later, so link tracking alone resolves 62% of those who acted.

Modelled on stated assumptions: 40% tap, 25% delayed search, 35% no action. · Why your influencer campaign has no data and how to fix it · Influencer and creator operations

The remaining 38% arrive as direct or organic traffic and need a discount code or a post-purchase survey to resolve to a creator.

Arithmetic on the same assumptions. · Why your influencer campaign has no data and how to fix it · Influencer and creator operations

Korant resolves a sale through 5 checks in a fixed priority order, with discount code ahead of cookie, which is what catches the delayed searchers a link misses.

Korant sale attribution priority order. · Why your influencer campaign has no data and how to fix it · Influencer and creator operations

A campaign launched without a unique tracking surface per creator cannot be reconstructed afterwards, so the setup has 1 viable moment: before the brief.

Derived from retrospective order data, which carries 0 recoverable per-creator identifiers. · Why your influencer campaign has no data and how to fix it · Influencer and creator operations

A 300-word script contains roughly 40 specifiable elements against 4 in a constraints brief, giving 10 times the surface area for a revision dispute.

Modelled by counting specifiable elements: wording, ordering, framing and timing instructions in a typical script. · The influencer brief that gets you usable content · Influencer and creator operations

At a 12% chance any specified element is missed or disputed, a script produces 4.8 expected revision triggers against 0.48 for a constraints brief.

Arithmetic on the same element counts at a stated 12% miss rate. · The influencer brief that gets you usable content · Influencer and creator operations

At 3 calendar days per revision round, that difference is 14.4 days against 1.4 for the same piece of content.

Arithmetic on the same assumptions at a stated 3 day round-trip. · The influencer brief that gets you usable content · Influencer and creator operations

ASCI's influencer guidelines, in force since 14 June 2021, place responsibility for disclosure on the advertiser as well as the influencer, so the label belongs in the brief.

ASCI, final Guidelines for Influencer Advertising in Digital Media, in force from 14 June 2021 · The influencer brief that gets you usable content · Influencer and creator operations

A Rs 6,00,000 launch budget at Rs 800 CAC buys 750 initial customers whether it is spent in one city or six, so density is the only variable that differs.

Modelled on stated assumptions: Rs 6,00,000 launch budget, Rs 800 customer acquisition cost. · Launching in one city at a time, on purpose · Hyperlocal and geo marketing

At a 30% raw referral rate, concentrating a launch lifts the effective rate from 9.0% to 25.5% and the final cohort from 824 to 1,007 customers.

Modelled on stated assumptions: referral reachability of 0.30 when live in 6 of 20 target cities against 0.85 within a single city. · Launching in one city at a time, on purpose · Hyperlocal and geo marketing

That is 22.1% more customers from identical spend, produced entirely by where the customers are rather than how many were bought.

Arithmetic on the same modelled cascade. · Launching in one city at a time, on purpose · Hyperlocal and geo marketing

The referral cascade multiplier moves from 1.10x to 1.34x across that range, and reaches 1.67x at a 40% effective referral rate.

Arithmetic on a geometric referral cascade, total equals initial divided by one minus the effective rate. · Launching in one city at a time, on purpose · Hyperlocal and geo marketing

A modelled brand with 8,000 customers holds about 402 of them in its second-ranked postcode, which is 5.03% of the customer base in one delivery zone.

Modelled on a Zipf distribution, exponent 0.8, across 320 pincodes, 10,800 annual orders at 1.35 orders per customer. · Local scarcity: making 400 people feel like the only ones · Hyperlocal and geo marketing

A bounded offer to 402 customers needs 19.9 times the response rate of an open offer to all 8,000 to produce the same number of orders.

Arithmetic on the modelled customer distribution above. · Local scarcity: making 400 people feel like the only ones · Hyperlocal and geo marketing

Matching an open offer that converts at 2% would require a 39.8% response rate from the bounded audience, which is why local scarcity wins on efficiency rather than on volume.

Arithmetic on the modelled customer distribution above. · Local scarcity: making 400 people feel like the only ones · Hyperlocal and geo marketing

The top-ranked postcode in the same model holds 701 customers against 111 in the tenth, so the size of a bounded audience varies more than 6 times across a rotation list.

Modelled on a Zipf distribution, exponent 0.8, across 320 pincodes. · Local scarcity: making 400 people feel like the only ones · Hyperlocal and geo marketing

Managing a creator on spreadsheets costs a modelled 1.50 hours a month, of which 0.50 is computing attribution and payout.

Modelled on stated task assumptions across chasing, verification, attribution, payment and relationship work. · Managing 200 creators without a spreadsheet graveyard · Influencer and creator operations

Adding per-creator tracking links cuts that to 0.85 hours and a full state machine with automated payout cuts it to 0.50.

Modelled on the same task breakdown at three tooling levels. · Managing 200 creators without a spreadsheet graveyard · Influencer and creator operations

At 80 hours of creator-facing time a month, that moves span of control from 53 creators per manager to 160, a factor of 3.

Arithmetic on the same hourly figures against 80 available hours. · Managing 200 creators without a spreadsheet graveyard · Influencer and creator operations

Cost per creator per month falls from Rs 1,200 to Rs 400 at Rs 800 an hour, before any change in what the creators themselves are paid.

Arithmetic on the same hourly figures at a stated Rs 800 hourly rate. · Managing 200 creators without a spreadsheet graveyard · Influencer and creator operations

Postcode appears in 0 of Shopify's prebuilt regional sales reports; the native Total Sales by Billing Location report groups by billing country and region instead.

Shopify Help Center, sales reports · A marketing channel nobody has ever touched · Hyperlocal and geo marketing

Every Shopify order with a shipping address carries a postcode, which makes pincode-level order data 100% available to D2C brands and never a data acquisition problem.

Shopify order shipping address fields. · A marketing channel nobody has ever touched · Hyperlocal and geo marketing

In a modelled 320-pincode order map, 41% of a 900-order month lands in just the top 20 pincodes.

Modelled on a Zipf distribution, exponent 0.8, across 320 pincodes and 900 monthly orders. · A marketing channel nobody has ever touched · Hyperlocal and geo marketing

An Indian PIN code's first 3 digits identify the sorting district, which gives a usable aggregation level when one pincode is too thin to test on.

India Post PIN code structure. · A marketing channel nobody has ever touched · Hyperlocal and geo marketing

On modelled Indian rate cards a nano creator delivers 1.6 attributed orders at a total cost of Rs 2,700, giving a CAC of Rs 1,688.

Modelled on stated assumptions: 5,000 followers, 20% reach, 4% click-through, 4% conversion, Rs 1,500 fee, 1.5 hours of management at Rs 800. · Nano-influencers in India: the unit economics · Influencer and creator operations

The same model gives a macro creator a CAC of Rs 10,356, which is 6.1 times the nano figure for 9 attributed orders.

Modelled on stated assumptions: 300,000 followers, 6% reach, Rs 90,000 fee, 4 hours of management. · Nano-influencers in India: the unit economics · Influencer and creator operations

Reaching 500 orders a month through nano creators requires roughly 312 of them and 469 hours of management, which is 2.9 person-months.

Arithmetic on the same assumptions at 1.6 orders and 1.5 management hours per creator. · Nano-influencers in India: the unit economics · Influencer and creator operations

At a working limit of 40 active creators per manager, that programme needs 7.8 managers, which is where nano economics actually fail.

Arithmetic on the same assumptions at a stated span of control. · Nano-influencers in India: the unit economics · Influencer and creator operations

Detecting a 20% relative change in order conversion at a 2.5% baseline needs 15,288 sessions per arm, or 25 days at 36,000 monthly sessions.

Two-proportion sample size at 80% power and 5% two-sided significance. · Nine D2C experiments you can run this week · D2C growth experiments

The same 20% change in checkout completion at a 30% baseline needs 915 sessions per arm, 16.7 times fewer, or 2 days.

Two-proportion sample size at 80% power and 5% two-sided significance. · Nine D2C experiments you can run this week · D2C growth experiments

Halving the target to a 10% relative change quadruples the requirement, taking order conversion from 25 days to 102.

Arithmetic on the same formula, where sample size scales with the inverse square of the effect. · Nine D2C experiments you can run this week · D2C growth experiments

Testing high-baseline metrics rather than order conversion is the single change that makes D2C experimentation feasible below 50,000 sessions a month.

Derived from the 16.7 times difference in sample requirement between the two baselines. · Nine D2C experiments you can run this week · D2C growth experiments

A modelled one-off post delivers 3.24 of its orders in week 1 and 4.98 in total, with almost nothing arriving after week 12.

Modelled on stated assumptions: 3.6 attributed orders, 90% in week one, 65% weekly decay thereafter. · Turning a one-off collab into a standing revenue line · Influencer and creator operations

A standing link adding 0.25 orders a week reaches 14.98 orders over 52 weeks, which is 3.0 times the one-off total.

Arithmetic on the same assumptions with a standing link from week 12. · Turning a one-off collab into a standing revenue line · Influencer and creator operations

With chained referral at a 15% rate, the same standing arrangement reaches 17.6 orders, or 3.5 times the one-off.

Arithmetic on a geometric referral cascade, total equals base divided by one minus the referral rate. · Turning a one-off collab into a standing revenue line · Influencer and creator operations

The standing arrangement adds 0 to the creator fee, since the tail is paid on commission against orders that would not otherwise exist.

Derived from a base-plus-commission structure with no additional fixed fee. · Turning a one-off collab into a standing revenue line · Influencer and creator operations

A modelled 20,000-follower creator's post produces 3.6 attributed orders, making a 15% commission-only deal worth Rs 972 against a Rs 8,000 flat fee ask.

Modelled on stated assumptions: 15% reach, 3% click-through, 4% conversion, Rs 1,800 average order value. · Paying influencers on performance without insulting them · Influencer and creator operations

Matching that flat fee on commission alone would require a 123% commission rate, so commission-only cannot be equivalent at typical creator volumes.

Arithmetic on the same assumptions: Rs 8,000 divided by Rs 6,480 of attributed revenue. · Paying influencers on performance without insulting them · Influencer and creator operations

A Rs 5,000 base plus 20% uncapped commission pays Rs 6,296 at expected performance and Rs 11,480 if the post performs 5 times better.

Arithmetic on the same assumptions. · Paying influencers on performance without insulting them · Influencer and creator operations

Under commission-only the creator carries 100% of the performance risk while the brand carries none, which is the part that reads as an insult.

Derived from the structure of a commission-only agreement. · Paying influencers on performance without insulting them · Influencer and creator operations

A Shopify order export carries dozens of columns and the pincode pivot needs 4 of them, plus the order identifier for deduplication.

Shopify order CSV export field list. · The pincode data you already have and have never exported · Hyperlocal and geo marketing

Shopify's prebuilt regional sales report groups by billing country and region, so 0 default views rank revenue by shipping postcode.

Shopify Help Center, sales reports · The pincode data you already have and have never exported · Hyperlocal and geo marketing

Shopify's native order CSV export includes 0 line item property columns, so any per-item custom data has to come from the API or a third-party export app.

Shopify Community, order export CSV fields · The pincode data you already have and have never exported · Hyperlocal and geo marketing

Joining the three files on the pincode key yields 6 measures per postcode, none of which appears in any one of the source systems.

Derived from the shared pincode key across order, serviceability and invoice files. · The pincode data you already have and have never exported · Hyperlocal and geo marketing

Roughly 8 in 10 pincodes in a modelled D2C order map produce fewer than three orders a month, which is too thin to run any offer test against.

Modelled on a Zipf distribution, exponent 0.8, across 320 pincodes and 900 monthly orders. · Pincode marketing: the geo lever D2C brands ignore · Hyperlocal and geo marketing

A sitewide 10% discount spends 92.6% of its budget on orders that would have happened without it.

Modelled on stated assumptions: 8% incremental lift, Rs 1,800 average order value, 900 orders. · Pincode marketing: the geo lever D2C brands ignore · Hyperlocal and geo marketing

Gating that same 10% discount to the top 20 pincodes changes cost per incremental order by 0, because it shrinks the subsidy base and the lift in equal proportion.

Modelled on the same assumptions, comparing 900 sitewide orders against 371 gated orders. · Pincode marketing: the geo lever D2C brands ignore · Hyperlocal and geo marketing

India's PIN system runs on 9 postal zones set by the first digit, 8 of them geographic and one reserved for the Army Postal Service.

India Post PIN code structure. · Pincode marketing: the geo lever D2C brands ignore · Hyperlocal and geo marketing

A personal WhatsApp message to 100 dormant affiliates has to reactivate about 6 of them to break even, against 0.02 for the same message sent as a bulk email.

Modelled on stated assumptions: Rs 800 per hour of operator time, 4 minutes per message, Rs 900 net contribution per reactivated affiliate. · Reactivating dormant affiliates without begging · Affiliate marketing

Korant resolves a sale through 5 checks in a fixed priority order, so an affiliate whose discount code is unmapped falls out of the first two before any cookie is read.

Korant sale attribution priority order. · Reactivating dormant affiliates without begging · Affiliate marketing

An affiliate who posts once a quarter sits outside a 30-day last-touch window for roughly 60 of every 90 days, so their sales resolve to whatever touched the buyer last.

Arithmetic on a 30-day last-touch cookie window. · Reactivating dormant affiliates without begging · Affiliate marketing

At a configured chain depth of 3, reactivating one dormant sharer in SB&R can pay coins to 3 people upstream who did nothing new that month.

SB&R chain payout behaviour at a depth of three. · Reactivating dormant affiliates without begging · Affiliate marketing

Of a modelled 1,000 readers, 60 click through and 19 eventually buy, but only 4 buy on that first visit.

Modelled on stated assumptions: 6% click-through, 6.7% same-visit conversion, 25% delayed conversion among clickers. · Publication and PR links: measuring what SEO teams can't · Influencer and creator operations

Last-touch attribution therefore credits 21% of what the publication produced among its own clickers, an undercount of 4.75 times.

Arithmetic on the same assumptions. · Publication and PR links: measuring what SEO teams can't · Influencer and creator operations

A further 25 readers who never clicked go on to buy after searching, so total contribution is 44 against the 4 last-touch sees, an 11 times undercount.

Arithmetic on the same assumptions. · Publication and PR links: measuring what SEO teams can't · Influencer and creator operations

Korant places cookie to publication 5th and last in its resolution order, which is correct for a channel that introduces rather than closes.

Korant sale attribution priority order. · Publication and PR links: measuring what SEO teams can't · Influencer and creator operations

A standard national shipping threshold plus a cash on delivery fee already produces a 7.2% spread in effective price between a metro core and a remote pincode.

Modelled on stated assumptions: Rs 1,800 list price, Rs 0 to Rs 99 shipping charged, Rs 30 COD fee in remote zones. · Regional pricing without a regional pricing team · Hyperlocal and geo marketing

Adding a 24.2% zone-gated discount in the metro widens that effective price spread from 7.2% to 41.4%.

Modelled on the same assumptions, with the discount applied only to metro core orders. · Regional pricing without a regional pricing team · Hyperlocal and geo marketing

Under that structure the remote customer pays 7.2% above list while the metro customer pays 24.2% below it, on identical products.

Modelled on stated assumptions: Rs 1,800 list price, zone-gated discount at 24.2%. · Regional pricing without a regional pricing team · Hyperlocal and geo marketing

Running regional pricing this way requires 0 changes to product prices, because the variation sits entirely in shipping, fees and discounts.

Derived from the shipping and discount mechanics described. · Regional pricing without a regional pricing team · Hyperlocal and geo marketing

Rotating one live zone per week across a 20 week cycle discounts 2.06% of orders, against 100% for an always-on offer.

Modelled on a Zipf distribution, exponent 0.8, across 320 pincodes at 900 monthly orders. · Rotating offers: one discount, many audiences, no code · Urgency, scarcity and flash

A 25% rotating discount on a 20 week cycle costs Rs 1,930 a week, which is 5.2% of the Rs 37,413 a week an always-on 10% discount costs.

Modelled on stated assumptions: Rs 1,800 average order value, 900 monthly orders, 208 weekly orders. · Rotating offers: one discount, many audiences, no code · Urgency, scarcity and flash

Shortening the cycle to 5 weeks raises the discounted share to 4.55% and the weekly cost to 11.4% of always-on, so cycle length is the budget dial.

Modelled on the same distribution and assumptions. · Rotating offers: one discount, many audiences, no code · Urgency, scarcity and flash

At a 20 week cycle an unchanged always-on budget would fund a 485% discount, which means the correct response is to cut the budget rather than deepen the offer.

Arithmetic on the same modelled weekly spend against the discounted order volume. · Rotating offers: one discount, many audiences, no code · Urgency, scarcity and flash

India's CCPA notified its dark patterns guidelines on 30 November 2023, listing 13 specified practices with false urgency first among them.

PIB, Government of India press release on the dark patterns advisory · Scarcity that's true: how to run urgency you can defend · Urgency, scarcity and flash

The guidelines give 2 illustrations of false urgency: faking a product's popularity, and falsely indicating that quantities are more limited than they actually are.

Guidelines for Prevention and Regulation of Dark Patterns, 2023, Annexure I · Scarcity that's true: how to run urgency you can defend · Urgency, scarcity and flash

Failure to comply with a CCPA order is punishable with imprisonment of up to 6 months or a fine of up to Rs 20 lakh, or both.

Guidelines for Prevention and Regulation of Dark Patterns, 2023, penalties under the Consumer Protection Act, 2019 · Scarcity that's true: how to run urgency you can defend · Urgency, scarcity and flash

On 5 June 2025 the CCPA directed e-commerce platforms to self-audit for dark patterns within 3 months and declare compliance.

PIB, Government of India press release on the dark patterns advisory · Scarcity that's true: how to run urgency you can defend · Urgency, scarcity and flash

A standard Indian courier serviceability export carries 9 fields per pincode, of which 7 change a marketing decision and 2 are purely operational.

Composite of serviceability exports from three Indian 3PL providers. · Serviceability data is a marketing asset, not an ops file · Hyperlocal and geo marketing

If a courier serves 5,000 pincodes and a brand has orders in 320, then 93.6% of the serviceable map has never produced a single order.

Arithmetic on stated assumptions: 5,000 serviceable pincodes against a 320-pincode order map. · Serviceability data is a marketing asset, not an ops file · Hyperlocal and geo marketing

Joining the serviceability file to an order export takes 1 lookup on the pincode column and outputs the serviceable-but-unsold list, which is the real expansion shortlist.

Derived from the shared pincode key in both files. · Serviceability data is a marketing asset, not an ops file · Hyperlocal and geo marketing

The expected transit time field converts 1 national shipping promise into a per-pincode promise at zero additional data cost.

Derived from the TAT column present in standard serviceability exports. · Serviceability data is a marketing asset, not an ops file · Hyperlocal and geo marketing

At a 25% second-order rate and 55% subsequent repeat, modelled lifetime contribution is Rs 980, which is the maximum CAC the brand can afford at a one-times payback.

Modelled on expected orders equal to one plus second-order rate over one minus repeat rate, at Rs 630 contribution per order. · The unlock nobody talks about: your second order · D2C growth experiments

Raising the second-order rate to 45% lifts lifetime contribution to Rs 1,260, a 29% higher CAC ceiling from the same product and price.

Arithmetic on the same formula. · The unlock nobody talks about: your second order · D2C growth experiments

A strong 17% CAC reduction from Rs 1,200 to Rs 1,000 gains Rs 200 per customer once, against Rs 280 per customer from the second-order change.

Arithmetic comparing the two levers at the same modelled parameters. · The unlock nobody talks about: your second order · D2C growth experiments

The CAC gain applies to customers you already buy, while the second-order gain also raises the ceiling, making channels priced above Rs 980 newly viable.

Derived from the difference between a cost reduction and a ceiling increase. · The unlock nobody talks about: your second order · D2C growth experiments

A modelled six month dark store commits Rs 20.9 lakh in fitout and running costs before any inventory, against Rs 2.77 lakh for a zone test across 3 cities.

Modelled on stated assumptions: 2,000 sq ft at Rs 60 per sq ft, 3 staff at Rs 25,000, Rs 8 lakh fitout, 4 discount windows per city. · Testing new cities with a discount instead of a warehouse · Hyperlocal and geo marketing

Including Rs 15 lakh of inventory tied up in the facility, the dark store costs 13 times the zone test for the same expansion decision.

Modelled on stated assumptions: Rs 20.9 lakh committed plus Rs 15 lakh inventory, against Rs 2.77 lakh of testing. · Testing new cities with a discount instead of a warehouse · Hyperlocal and geo marketing

A zone test is 13.3% of the committed cost of a dark store and can be stopped after any single 7 day window with no residual liability.

Modelled on the same assumptions, comparing committed spend rather than total spend. · Testing new cities with a discount instead of a warehouse · Hyperlocal and geo marketing

A discount-led city test measures price-elastic demand only, so it overstates baseline demand by the full lift the discount produced, typically 8% to 40%.

Modelled lift range used across this cluster's discount economics. · Testing new cities with a discount instead of a warehouse · Hyperlocal and geo marketing

In a 20 zone rotation with one live per week, any given zone is live 5% of the time, so 95% of referrals happen outside a live window.

Arithmetic on a 20 zone rotation cycle with one zone live per week. · Time-boxed offers and the two clocks you must never mix · Urgency, scarcity and flash

Tying the reward clock to the offer clock therefore pays nothing on 95% of referrals, including ones made minutes before the window closed.

Derived from the same rotation arithmetic. · Time-boxed offers and the two clocks you must never mix · Urgency, scarcity and flash

A 30 day referral link validity covers 4.3 weeks, which is 21% of a 20 week rotation cycle, so the reward clock has to be set independently of the cycle.

Arithmetic on a 30 day link validity against a 20 week rotation cycle. · Time-boxed offers and the two clocks you must never mix · Urgency, scarcity and flash

The offer clock is evaluated at checkout and the reward clock at order creation, which are 2 distinct moments in an order lifecycle.

Derived from Shopify discount evaluation and order creation sequencing. · Time-boxed offers and the two clocks you must never mix · Urgency, scarcity and flash

On a modelled traffic mix of 70% first visits and 30% repeat visits, repeat visitors produce 56.2% of orders.

Modelled on stated assumptions: 1.8% first-visit conversion against 5.4% repeat-visit conversion. · Urgency that survives a second visit · Urgency, scarcity and flash

A repeat visitor averaging 4 visits has a 47.8% chance of having caught a resettable element, which is 26.9% of all orders coming from an audience that ran the test.

Modelled on a 15% per-exposure detection probability applied to the same traffic mix. · Urgency that survives a second visit · Urgency, scarcity and flash

At 10 visits that rises to 45.2% of all orders coming from shoppers who have already established that the urgency element is not real.

Modelled on the same detection probability and traffic mix. · Urgency that survives a second visit · Urgency, scarcity and flash

First-time visitors can detect 0 resettable elements, so the entire credibility risk sits with the segment producing most of the revenue.

Derived from the definition of a first visit. · Urgency that survives a second visit · Urgency, scarcity and flash

A referral link on the order confirmation reaches a modelled 96 people per 100 orders for 0.5 days of work, the best ratio of the 12 mechanics compared.

Modelled on stated assumptions: 100% of buyers see it, 12% share, each share reaching 8 people. · Viral D2C hacks to grow sales · D2C growth experiments

Raising chain depth from 1 to 3 reaches 153.6 people per 100 orders for 1 day of configuration, the highest absolute reach in the set.

Modelled on the same share rate with a 1.6x multiplier from upstream cascade. · Viral D2C hacks to grow sales · D2C growth experiments

A public referrer leaderboard takes 3 days to ship and returns 26.7 people reached per day of work, against 192.0 for the order confirmation link.

Modelled on stated assumptions: 4% participation reaching 20 people each. · Viral D2C hacks to grow sales · D2C growth experiments

The 4 fastest mechanics all attach to a moment that already happens, which is why they cost under a day and need no new traffic.

Derived from the effort ranking of the twelve mechanics compared. · Viral D2C hacks to grow sales · D2C growth experiments

Opening a new zone at a modelled Rs 15,000 seeding cost needs 23.8 orders at Rs 630 contribution each before it breaks even.

Modelled on stated assumptions: Rs 1,800 average order value, 35% contribution, Rs 15,000 of seeding per zone launch. · Building a waitlist by pincode · Hyperlocal and geo marketing

A 200-signup pincode waitlist needs 11.9% waitlist-to-order conversion to cover the zone opening cost, against 47.6% at 50 signups.

Arithmetic on the same assumptions, dividing 23.8 required orders by waitlist size. · Building a waitlist by pincode · Hyperlocal and geo marketing

At 400 signups the required conversion falls to 6.0%, which is why 200 is the lowest defensible unlock threshold for a pincode.

Arithmetic on the same assumptions. · Building a waitlist by pincode · Hyperlocal and geo marketing

Every blocked checkout in an unserviceable pincode is 1 intent signal, and stores without a waitlist capture 0 of them.

Derived from standard serviceability-blocking behaviour at checkout. · Building a waitlist by pincode · Hyperlocal and geo marketing

A creator campaign has 7 sequential failure points, and a zero-order result looks the same at every one of them from the outside.

Derived from the campaign chain: distribution, link, landing, offer availability, attribution, audience fit, creative. · When a creator campaign flops, what to check first · Influencer and creator operations

Reach, clicks and orders are 3 numbers that localise the failure to one link in the chain in under an hour.

Derived from the same chain, since each break produces a distinct pattern across the three figures. · When a creator campaign flops, what to check first · Influencer and creator operations

A healthy modelled post runs 3,000 reach to 90 clicks to 3.6 orders, so a 3% click rate and a 4% conversion are the reference points to compare against.

Modelled on stated assumptions: 15% reach on 20,000 followers, 3% click-through, 4% conversion. · When a creator campaign flops, what to check first · Influencer and creator operations

Clicks landing normally with 0 attributed orders is the signature of a tracking failure rather than a creative one.

Derived from the difference between traffic records and attribution records. · When a creator campaign flops, what to check first · Influencer and creator operations

A modelled share into a group of 8, with a 10% onward forward rate, reaches 31.6 people across generations from a single original share.

Arithmetic on a geometric fan-out with a group size of 8 and a 10% forward rate, giving a decay factor of 0.8. · WhatsApp as a growth channel for Indian D2C · D2C growth experiments

Of those 31.6 people, 0 arrive carrying a referrer header, while 100% carry whatever identifier is inside the URL itself.

Derived from how links opened from a messaging app present to a web server. · WhatsApp as a growth channel for Indian D2C · D2C growth experiments

Every generation reports as the same source, so 32 visits from one slug cannot be distinguished from 32 separate original shares.

Derived from the absence of generation data in a forwarded link. · WhatsApp as a growth channel for Indian D2C · D2C growth experiments

A redirect slug survives forwarding where a referrer does not, which makes the identifier's position in the URL the 1 decision that determines visibility.

Derived from URL persistence across message forwarding. · WhatsApp as a growth channel for Indian D2C · D2C growth experiments

At a Rs 1,800 AOV and 35% gross margin, contribution is Rs 630, so a CAC equal to AOV is 2.86 times the contribution of a single order.

Arithmetic on stated assumptions: Rs 1,800 average order value at 35% gross margin. · What to do when your CAC crosses your AOV · D2C growth experiments

A CAC of Rs 1,800 repays in 2.86 orders, which is 257 days at a 90 day repurchase cycle, against 171 days for a CAC of Rs 1,200.

Arithmetic on the same assumptions at a stated 90 day cycle. · What to do when your CAC crosses your AOV · D2C growth experiments

At a 20% second-order rate and 40% subsequent repeat, lifetime contribution is Rs 840, so a Rs 1,800 CAC loses Rs 960 per customer.

Modelled on expected orders equal to one plus second-order rate divided by one minus repeat rate. · What to do when your CAC crosses your AOV · D2C growth experiments

At a 65% second-order rate and 70% repeat, the same Rs 1,800 CAC returns Rs 195 per customer, so repeat behaviour decides whether the ratio is survivable.

Modelled on the same expected-orders formula. · What to do when your CAC crosses your AOV · D2C growth experiments

A waitlist capturing 14% of an 8,000-customer base converts at a modelled 18% on launch day, producing 202 orders with no media spend.

Modelled on stated assumptions: 8,000 all-time customers, 14% signup, 18% launch-day conversion. · Zero-budget launches that actually moved numbers · D2C growth experiments

Zone-gated early access across the top 6 pincodes reaches 1,984 customers and converts at 9%, producing 179 orders.

Modelled on a Zipf order distribution where the top 6 pincodes hold 24.8% of customers. · Zero-budget launches that actually moved numbers · D2C growth experiments

Chained invites at 2 per customer reach 640 people at 8% participation and convert at 22%, producing 141 orders.

Modelled on stated assumptions: 8% of customers use an invite, 22% recipient conversion. · Zero-budget launches that actually moved numbers · D2C growth experiments

Running all three does not produce 521 orders, because they share one pool; at 35% overlap the realistic figure is 339.

Arithmetic on the same customer base with a stated 35% overlap between mechanics. · Zero-budget launches that actually moved numbers · D2C growth experiments

The budget that funds a 10% sitewide discount on 900 orders funds a 24.2% discount across the top 20 postcodes, which hold 371 of those orders.

Modelled on stated assumptions: Rs 1,800 average order value, Rs 162,000 discount budget, Zipf distribution with exponent 0.8 across 320 pincodes. · The zip-code discount that beats a sitewide sale · Hyperlocal and geo marketing

Gating a discount without deepening it changes cost per incremental order by 0, because it shrinks the subsidy base and the lift in equal proportion.

Modelled on stated assumptions: 8% incremental lift applied to both the sitewide and gated populations. · The zip-code discount that beats a sitewide sale · Hyperlocal and geo marketing

A concentrated discount in the top 20 postcodes has to generate a 19.5% conversion lift to match the 67 incremental orders a 10% sitewide discount produces.

Modelled on stated assumptions: 8% sitewide lift, Rs 162,000 budget held constant, 344 baseline orders in the gated zone. · The zip-code discount that beats a sitewide sale · Hyperlocal and geo marketing

Narrowing to the top 5 postcodes pushes the budget-equivalent depth to 44%, which is past the point most D2C gross margins survive.

Modelled on stated assumptions: Rs 162,000 budget across 205 monthly orders in the top 5 postcodes. · The zip-code discount that beats a sitewide sale · Hyperlocal and geo marketing

A modelled 76% of last-click affiliate sales had an earlier touch from a non-affiliate channel.

Modelled on stated assumptions: 1,000 affiliate-attributed orders scored against prior touches across paid, search, email and creator channels. · Affiliate attribution: last click is lying to you · Affiliate marketing

Sales attributed to a browser coupon extension had an earlier non-affiliate touch a modelled 89% of the time, against 52% for content-link affiliates.

Modelled on stated assumptions comparing interception behaviour against genuine referral traffic. · Affiliate attribution: last click is lying to you · Affiliate marketing

Only a modelled 24% of last-click affiliate sales had no earlier touch from any other channel.

Derived from the same 1,000-order breakdown. · Affiliate attribution: last click is lying to you · Affiliate marketing

A public discount code appears on aggregator sites within days, at which point it stops identifying the partner it was issued to.

Derived from the code leakage mechanic described in this article. · Affiliate attribution: last click is lying to you · Affiliate marketing

A fixed 5-step priority order resolves competing claims identically every month, which is what makes affiliate reporting comparable across periods.

Derived from the documented priority order described in the resolution section. · Affiliate attribution: last click is lying to you · Affiliate marketing

Minimum contribution margin required equals commission rate divided by 0.35, which puts a flat 10% commission's floor at 28.6% of order value.

Derived from the payout ceiling rule applied to commission structures. · The affiliate commission structures that survive a margin review · Affiliate marketing

A flat 15% commission needs contribution margin of 42.9% of order value, which most D2C catalogues do not have after order costs.

Modelled on stated assumptions: order costs of ₹83 on a ₹1,450 order reducing 45% gross margin to 39% contribution margin. · The affiliate commission structures that survive a margin review · Affiliate marketing

A three-level chain at 8%, 4% and 2% totals 14% and needs a 40% contribution margin floor, marginally above the modelled 39%.

Modelled on stated assumptions, applying the same ceiling rule to a cumulative chain payout. · The affiliate commission structures that survive a margin review · Affiliate marketing

A hybrid of 6% plus a ₹50 volume bonus averages a modelled 6.3% and needs only an 18% contribution margin floor.

Modelled on stated assumptions: a bonus paid to affiliates clearing ten sales a month, spread across all commissioned orders. · The affiliate commission structures that survive a margin review · Affiliate marketing

Per-lead commission is the only structure with unbounded fraud exposure, because a lead costs almost nothing to fabricate.

Derived from the cost-to-fake comparison across trigger events. · The affiliate commission structures that survive a margin review · Affiliate marketing

Coupon listicles produce a modelled 4.0 orders per thousand impressions against 2.8 for comparison posts, and 0.44 incremental orders against 1.62.

Modelled on stated assumptions: click-through and conversion rates by format, multiplied by modelled incrementality. · The affiliate content that actually converts · Affiliate marketing

Use-case demos model the highest incrementality of any published format at 64%, against 11% for coupon listicles.

Modelled on stated assumptions about audience intent at the moment each format is encountered. · The affiliate content that actually converts · Affiliate marketing

A one-to-one personal recommendation converts at a modelled 11% of clicks, against 3.4% for a best-of listicle.

Modelled on stated assumptions comparing addressed messages with published content. · The affiliate content that actually converts · Affiliate marketing

Comparison posts convert at a modelled 6.8% because the reader has narrowed to a shortlist and is looking for a tiebreaker.

Modelled on stated assumptions about reader intent in comparison content. · The affiliate content that actually converts · Affiliate marketing

Ranking formats by raw orders reverses the ranking by incremental orders almost completely.

Derived from the format comparison table in this article. · The affiliate content that actually converts · Affiliate marketing

A three-number dashboard models to 4.1 logins per affiliate per month against 1.3 for a dashboard with fifteen or more metrics.

Modelled on stated assumptions: login frequency falling as the answer to the primary question becomes harder to locate. · The affiliate dashboard your partners will actually log into · Affiliate marketing

Support tickets run at a modelled 3 per 100 affiliates per month with three numbers, against 12 with a complex dashboard and 24 with no dashboard at all.

Modelled on stated assumptions: tickets driven by partners unable to answer their own question. · The affiliate dashboard your partners will actually log into · Affiliate marketing

Pending is the number that prevents most tickets, because it explains why a known sale has not become money yet.

Derived from the ticket categories described in this article. · The affiliate dashboard your partners will actually log into · Affiliate marketing

Every metric added to the load view moves the primary answer further down the page, which is why complexity reduces logins rather than deepening engagement.

Derived from the login frequency comparison across dashboard designs. · The affiliate dashboard your partners will actually log into · Affiliate marketing

A dashboard that cannot be reconciled to the brand's own ledger creates disputes that no amount of interface work resolves.

Derived from the reconciliation requirement described in this article. · The affiliate dashboard your partners will actually log into · Affiliate marketing

Manual affiliate management models to 165 hours a week at 1,000 affiliates against 4 hours with self-serve signup and automated payouts.

Modelled on stated assumptions: roughly 0.16 hours per affiliate per week manually, against a fixed base plus exception review. · Affiliate marketing for D2C brands with no affiliate manager · Affiliate marketing

At 100 affiliates the modelled gap is 19 hours a week against 1.5, which is the point where most brands hire someone.

Modelled on stated assumptions, from the same workload comparison. · Affiliate marketing for D2C brands with no affiliate manager · Affiliate marketing

Exception-based review touches a modelled 3% of affiliates at about eight minutes each, which is 4 hours a week at 1,000 affiliates.

Modelled on stated assumptions about the share of accounts triggering a review rule. · Affiliate marketing for D2C brands with no affiliate manager · Affiliate marketing

Store credit payouts remove verification, payout rails and reconciliation, which is the largest single component of manual workload.

Derived from the workload breakdown by task in this article. · Affiliate marketing for D2C brands with no affiliate manager · Affiliate marketing

4 review rules cover most exceptions worth a human: velocity, refund rate, single-buyer concentration, and traffic with no engagement.

Derived from the exception rules set out in this article. · Affiliate marketing for D2C brands with no affiliate manager · Affiliate marketing

Affiliates who make a first sale within 48 hours model to 71% still active at six months, against 14% for those taking over 30 days.

Modelled on stated assumptions: six-month retention by time-to-first-sale cohort. · Affiliate onboarding: the first 48 hours decide everything · Affiliate marketing

A modelled 38% of affiliates never make a sale at all, and 4% of those are still active at six months.

Modelled on stated assumptions across the same cohort breakdown. · Affiliate onboarding: the first 48 hours decide everything · Affiliate marketing

Weighted across all cohorts, a modelled 27.7% of recruited affiliates are still active at six months.

Derived from the cohort shares and retention rates in the table. · Affiliate onboarding: the first 48 hours decide everything · Affiliate marketing

Time from intent to first share falls from a modelled 6.2 days under an application process to under a minute when the link already exists.

Modelled on stated assumptions about review turnaround and approval notification. · Affiliate onboarding: the first 48 hours decide everything · Affiliate marketing

One specific first action beats a list of options, because a partner given 6 ways to start picks none of them.

Derived from the onboarding step design described in this article. · Affiliate onboarding: the first 48 hours decide everything · Affiliate marketing

Commission and brokerage payments attract TDS under Section 194H of the Income Tax Act, subject to a threshold per financial year.

Income Tax Act, 1961, Section 194H. Rates and thresholds have been amended repeatedly, most recently through the Finance Acts of 2024 and 2025. Verify current figures with your CA. · Affiliate payouts in India: TDS, GST, and getting it right · Affiliate marketing

Benefits or perquisites provided in kind rather than in cash are addressed separately under Section 194R, which is the section to raise when rewards are store credit.

Income Tax Act, 1961, Section 194R, inserted by the Finance Act 2022. Applicability to store credit is a question for your CA rather than a settled position. · Affiliate payouts in India: TDS, GST, and getting it right · Affiliate marketing

An affiliate supplying services to a brand is making a taxable supply under GST, with the standard rate for such services at 18%.

Central Goods and Services Tax Act, 2017. Whether GST applies depends on the affiliate's registration status and turnover. Verify with your CA. · Affiliate payouts in India: TDS, GST, and getting it right · Affiliate marketing

There are 4 deduction and documentation points in a cash payout workflow and 1 in a store credit workflow.

Derived from the workflow comparison in this article. · Affiliate payouts in India: TDS, GST, and getting it right · Affiliate marketing

Unregistered affiliates below the GST registration threshold do not charge GST, which is why most customer-affiliate programs never generate an invoice.

Central Goods and Services Tax Act, 2017, registration provisions. Thresholds vary by state category and supply type. Verify with your CA. · Affiliate payouts in India: TDS, GST, and getting it right · Affiliate marketing

7 clauses cover the failures that produce most affiliate losses, and each one maps to a single identifiable event.

Derived from the clause table in this article. · Affiliate program terms that prevent your worst month · Affiliate marketing

Brand bidding, clause 1 of 7, is the most expensive omission, because an affiliate competing in your own brand-term auction raises your paid search costs while collecting commission on demand you created.

Derived from the failure described for the brand bidding clause. · Affiliate program terms that prevent your worst month · Affiliate marketing

Clause 4, a cooling-off period holding commission until the return window closes, removes refund farming without any detection logic.

Derived from the cooling-off clause and its failure mode. · Affiliate program terms that prevent your worst month · Affiliate marketing

Clause 5 exists because a code that leaks to aggregator sites stops identifying the partner it was issued to, which is why confidentiality is a clause rather than a request.

Derived from the code confidentiality clause in this article. · Affiliate program terms that prevent your worst month · Affiliate marketing

Clause 6, publishing the attribution priority order, converts disputes from arguments about fairness into questions about a written rule.

Derived from the attribution priority clause. · Affiliate program terms that prevent your worst month · Affiliate marketing

About 35% of credit moves away from the channel that created demand when conversions land outside the attribution window.

Modelled on stated assumptions: 7-day window, 35% of orders converting after day seven, single-touch crediting. · Attribution window: what it is and how window length changes credit · Attribution and measurement

Korant resolves ambiguous sales through a fixed 5-step priority order rather than a single window, starting with discount code to influencer and ending with cookie to publication.

Korant attribution resolution order, product documentation. · Attribution window: what it is and how window length changes credit · Attribution and measurement

Summed platform-reported revenue can exceed a store's real revenue, because 2 platforms with different windows can each claim the same order without either being wrong.

Consequence of platforms attributing independently, derived rather than measured. · Attribution window: what it is and how window length changes credit · Attribution and measurement

A 30-day window credits touchpoints up to a month before purchase, which flatters channels that reach people already intending to buy.

Definitional consequence of a 30-day window under single-touch crediting. · Attribution window: what it is and how window length changes credit · Attribution and measurement

A Shopify store can have a maximum of 25 active automatic discounts at one time, and that total includes discounts created by installed apps.

Shopify Help Center, automatic discounts · Automatic discount on Shopify: what it is and how it differs from code · Shopify stack and extensions

Shoppers can apply at most 5 product or order discount codes to a single order, a per-order limit that does not constrain automatic discounts in the same way.

Shopify Help Center, discount combinations and limits · Automatic discount on Shopify: what it is and how it differs from code · Shopify stack and extensions

Active automatic app-based discounts were capped at 5 before the limit was raised, which is why discount apps could previously hold only a handful of concurrent offers.

Shopify developer changelog, increased limits for automatic function-based discounts · Automatic discount on Shopify: what it is and how it differs from code · Shopify stack and extensions

A code intended for a 200-customer winback segment has no upper bound on redemptions once it is posted publicly, because the string works for anyone who types it.

Modelled on stated assumptions: 200-customer target segment, code shared to a coupon aggregator, no usage limit configured. · Automatic discount on Shopify: what it is and how it differs from code · Shopify stack and extensions

A fully populated three-level chain costs 17% of order value on every referred order, before any reward cap is applied.

Modelled on stated assumptions: 10%, 5% and 2% of order value paid at levels one, two and three, no cap, all levels populated. · Chained referral: how one order pays more than one person · Referral and word of mouth

Level three of a chain fires on roughly 16% of referred orders once chains that are not fully populated are accounted for.

Modelled on stated assumptions: 40% independent probability that each upstream level exists, three configured levels. · Chained referral: how one order pays more than one person · Referral and word of mouth

Shopify permits a maximum of 5 product or order discount codes on a single order, which is why cascading rewards are banked as credit rather than issued as stacked codes.

Shopify Help Center, discount combinations and limits · Chained referral: how one order pays more than one person · Referral and word of mouth

A three-level chain produces up to 3 separate ledger entries from one referred order, one per populated level.

Structural property of chained payouts, stated for a three-level configuration. · Chained referral: how one order pays more than one person · Referral and word of mouth

A three-level chain paying 8%, 4% and 2% costs ₹203 on a ₹1,450 order, leaving ₹449 of the ₹652 gross margin intact.

Modelled on stated assumptions: ₹1,450 AOV, 45% gross margin, coins valued at face rather than at redeemed cost. · Chained referrals: paying three people for one sale, profitably · Referral and word of mouth

Moving from one level to two models to 18 more referred orders a month for ₹6,612 of extra coins, a net gain of ₹5,133.

Modelled on stated assumptions: 60 referred orders at depth one rising to 78 at depth two, on a 1,000-order base. · Chained referrals: paying three people for one sale, profitably · Referral and word of mouth

Moving from depth four to depth five models to a net loss of ₹240 a month, which is where the chain stops paying for itself.

Modelled on stated assumptions: one additional referred order against an extra 0.5% payout applied across 93 referred orders. · Chained referrals: paying three people for one sale, profitably · Referral and word of mouth

At a modelled 68% redemption rate, ₹203 of coins issued costs about ₹138 in realised margin.

Modelled on stated assumptions: unredeemed coins expire or sit unspent and never reach the P&L. · Chained referrals: paying three people for one sale, profitably · Referral and word of mouth

Total payout per order equals what one participant can earn across three generations of their own chain: ₹203.

Modelled on stated assumptions, derived from the A to E worked example at 8%, 4% and 2%. · Chained referrals: paying three people for one sale, profitably · Referral and word of mouth

The nine tweaks model to about a 12% AOV lift in total, but they overlap, so on a ₹1,450 base a realistic plan is roughly ₹85 more per order.

Modelled on stated assumptions, with the working shown per tweak and the overlap explained in the effort versus lift section. · Nine Shopify checkout tweaks that lift AOV without a redesign · Shopify stack and extensions

Raising a free shipping threshold from ₹999 to ₹1,399 models to about ₹24 more per order before abandonment.

Modelled on stated assumptions: ₹1,450 AOV, ₹70 shipping recovered, 18% of orders inside the band, 50% of them topping up. · Nine Shopify checkout tweaks that lift AOV without a redesign · Shopify stack and extensions

A quantity break that starts at two units models to about ₹35 more per order on a single-unit-heavy catalogue.

Modelled on stated assumptions: 68% single-unit orders, ₹800 unit price, 10% off at two units, 8% of eligible orders converting. · Nine Shopify checkout tweaks that lift AOV without a redesign · Shopify stack and extensions

A 10% offer gated to one high-density pincode zone models at about 2.5x less discount spend per incremental order than the same offer run sitewide.

Modelled on stated assumptions: 1,000 orders at ₹1,450, 8% incremental lift sitewide, 12% of orders inside the gated zone with a 20% lift there. · Nine Shopify checkout tweaks that lift AOV without a redesign · Shopify stack and extensions

Capping store credit redemption at 15% of order value models to about ₹15 more per order across the whole book.

Modelled on stated assumptions: 20% of orders carry a redeemable balance averaging ₹300, and 30% of those shoppers raise the basket by ₹250. · Nine Shopify checkout tweaks that lift AOV without a redesign · Shopify stack and extensions

A one-to-one coin denomination is valued correctly by a modelled 94% of shoppers against 23% for a scheme where one point equals ₹0.037.

Modelled comprehension test on stated assumptions: shoppers asked to state a balance's rupee value without using a calculator. · Coins, wallets, and points: what shoppers actually understand · Referral and word of mouth

Median time to answer is a modelled 2 seconds for one-to-one coins and 41 seconds for fractional points.

Modelled on stated assumptions about the arithmetic each scheme requires before an answer is possible. · Coins, wallets, and points: what shoppers actually understand · Referral and word of mouth

A modelled 38% of shoppers overestimate the value of a fractional points balance, because a four-digit number reads as a large reward.

Modelled on stated assumptions: 2,700 points presented without a conversion rate visible in the same view. · Coins, wallets, and points: what shoppers actually understand · Referral and word of mouth

Redemption rate tracks comprehension: a modelled 68% for one-to-one coins, 41% for fractional points, 33% for tiered percentage-back.

Modelled on stated assumptions linking the share of shoppers who can value a balance to the share who spend it. · Coins, wallets, and points: what shoppers actually understand · Referral and word of mouth

Fractional denominations were designed to make rewards feel larger and instead make them uncomputable, which is why they under-redeem.

Derived from the modelled comprehension and redemption figures in this article. · Coins, wallets, and points: what shoppers actually understand · Referral and word of mouth

Modelled incrementality runs from 8% for browser coupon extensions to 68% for customer referral links, a more than eightfold spread.

Modelled on stated assumptions: the share of attributed orders that would not have happened without the partner. · Coupon-code affiliates vs link affiliates: run both, differently · Affiliate marketing

A single flat commission across all partner types overpays extension traffic by roughly 8x relative to what its incrementality supports.

Derived from the incrementality spread applied to a uniform rate. · Coupon-code affiliates vs link affiliates: run both, differently · Affiliate marketing

Deal and coupon communities model at 22% incrementality against 55% for content and review sites.

Modelled on stated assumptions comparing discount-seeking audiences with research-led ones. · Coupon-code affiliates vs link affiliates: run both, differently · Affiliate marketing

A public code stops identifying the partner it was issued to within days, once it is republished on aggregator sites.

Derived from the code leakage mechanic described in this article. · Coupon-code affiliates vs link affiliates: run both, differently · Affiliate marketing

Commission scaled to incrementality gives a modelled range from 1% to 12% across the six partner types, against a typical flat 10%.

Derived from the recommended rates in the comparison table. · Coupon-code affiliates vs link affiliates: run both, differently · Affiliate marketing

Application-gated recruitment converts a modelled 1.1% of customers shown the ask into affiliates who ever share anything.

Modelled on stated assumptions: 8.4% start, 49% complete, 63% approved, 42% ever share, on 1,000 customers shown the ask. · Turning customers into affiliates without an application form · Affiliate marketing

Purchase-gated recruitment converts a modelled 5.1%, a 4.6x difference on the same audience and the same reward.

Modelled on stated assumptions: 11.8% engage with an existing link, 43% of those share. · Turning customers into affiliates without an application form · Affiliate marketing

Median time from intent to first share falls from a modelled 6.2 days under an application to under a minute when the link already exists.

Modelled on stated assumptions about review turnaround and approval notification. · Turning customers into affiliates without an application form · Affiliate marketing

An application funnel rejects a modelled 15 of every 41 completed applications, each one a customer told no by a brand they just paid.

Modelled on stated assumptions: a 63% approval rate applied to completed applications. · Turning customers into affiliates without an application form · Affiliate marketing

The purchase is a stronger filter than most application forms, because it requires spending money at full price rather than filling in fields.

Derived from the comparison of what each gate actually verifies. · Turning customers into affiliates without an application form · Affiliate marketing

Cookie stuffing, the 1st of 4 fraud types, shows up as attributed orders exceeding recorded landing page views for the same affiliate, which is arithmetically impossible for genuine traffic.

Derived from the detection signal described for cookie stuffing in this article. · Detecting affiliate fraud before you pay it out · Affiliate marketing

A modelled 3% of affiliate accounts trigger a review rule, at roughly eight minutes each, which is four hours a week at a thousand affiliates.

Modelled on stated assumptions about the share of accounts crossing a detection threshold. · Detecting affiliate fraud before you pay it out · Affiliate marketing

Holding commission until the return window closes removes 100% of refund farming, with no detection logic running.

Derived from the payout-hold rule described for refund farming. · Detecting affiliate fraud before you pay it out · Affiliate marketing

Brand bidding, 1 of the 4 fraud types, is detected from first touch rather than last, since the paid brand-term click is usually not the final touch before purchase.

Derived from the detection signal described for brand bidding. · Detecting affiliate fraud before you pay it out · Affiliate marketing

All 4 rules hold a payout rather than banning an account, because the false positive cost of a hold is a delay and the cost of a ban is a partner.

Derived from the payout-hold design described throughout. · Detecting affiliate fraud before you pay it out · Affiliate marketing

A metafield value above 10,000 bytes returns null inside a Shopify Function input query, even though the metafield itself stores far more.

Shopify developer changelog, Functions input limit updates · The five metafields every D2C store should be writing · Shopify stack and extensions

JSON metafield writes are capped at 128KB from API version 2026-04, with apps predating April 2026 grandfathered at the old 2MB limit.

Shopify developer changelog, reduced metafield value sizes · The five metafields every D2C store should be writing · Shopify stack and extensions

Most metafield types carry a 64KB size limit, which is six times what a Function can actually read.

Shopify developer documentation, metafield limits · The five metafields every D2C store should be writing · Shopify stack and extensions

Shopify Functions have 0 network access at runtime, which is the single constraint that makes metafields load-bearing.

Shopify Function APIs documentation · The five metafields every D2C store should be writing · Shopify stack and extensions

Rotating an offer once a day means 30 metafield writes a month instead of 900 checkout-time lookups.

Derived: one write per daily rotation across 30 days, against a 900 order month. · The five metafields every D2C store should be writing · Shopify stack and extensions

Five affiliates at 200 orders each mean any single departure removes 20% of referred volume overnight.

Derived from the concentration arithmetic of a five-partner program producing 1,000 orders. · Micro-affiliates: why 500 people at 2 sales beats 5 at 200 · Affiliate marketing

Modelled incrementality is 38% for macro affiliates against 66% for micro, because macro audiences overlap more heavily with existing paid targeting.

Modelled on stated assumptions: 41% audience overlap with paid social for macro partners against 9% for micro. · Micro-affiliates: why 500 people at 2 sales beats 5 at 200 · Affiliate marketing

The same 1,000 referred orders produce a modelled 380 incremental orders through macro partners and 660 through micro.

Derived from the modelled incrementality rates applied to identical order volume. · Micro-affiliates: why 500 people at 2 sales beats 5 at 200 · Affiliate marketing

Management runs a modelled 12.5 hours a week for five macro partners against 2.5 hours for five hundred micro ones.

Modelled on stated assumptions: negotiated relationships requiring individual attention against exception-based review at 3% of accounts. · Micro-affiliates: why 500 people at 2 sales beats 5 at 200 · Affiliate marketing

Micro programs remain power-law distributed, with a modelled top 12% producing half of referred revenue, but no single participant exceeds 2%.

Modelled on stated assumptions about revenue concentration within a five-hundred-affiliate cohort. · Micro-affiliates: why 500 people at 2 sales beats 5 at 200 · Affiliate marketing

India's average ecommerce return-to-origin rate is roughly 23%, and it varies sharply between delivery postcodes rather than spreading evenly across them.

GoKwik analysis across more than 180 million shoppers, reported 2026 · Pincode targeting vs city targeting: why the delivery address wins · Hyperlocal and geo marketing

A discount confined to a delivery zone holding 10% of order volume exposes about a tenth of the orders a sitewide offer at the same rate would discount.

Modelled on stated assumptions: target zone accounts for 10% of order volume, identical discount rate in both cases, no demand shift. · Pincode targeting vs city targeting: why the delivery address wins · Hyperlocal and geo marketing

Prepaid orders in India return undelivered at under 2%, so the payment mix inside a delivery zone changes its true fulfilment cost independently of demand.

GoKwik data reported across Indian logistics coverage, 2026 · Pincode targeting vs city targeting: why the delivery address wins · Hyperlocal and geo marketing

A Shopify discount function reads the delivery postcode from checkout-owned data at cart.deliveryGroups[0].deliveryAddress.zip rather than from a shopper-writable cart attribute.

Shopify Functions discount API input schema · Pincode targeting vs city targeting: why the delivery address wins · Hyperlocal and geo marketing

A fraudulent referrer identity costs nothing to create in an open referral program and one full order in a purchase-gated one, or ₹1,200 at a typical Indian D2C basket.

Modelled on stated assumptions: ₹1,200 average order value, one completed order required before a referral link exists. · Purchase-gated referral: what it is and how the gate works · Referral and word of mouth

Purchase gating caps the referrer pool at the customer list, so a brand with 1,000 customers has at most 1,000 possible referrers no matter how much traffic the store gets.

Definitional limit of purchase gating, stated on a 1,000-customer base. · Purchase-gated referral: what it is and how the gate works · Referral and word of mouth

Shopify allows a maximum of 25 active automatic discounts per store, including app-created ones, which caps how many reward tiers can redeem at checkout at the same time.

Shopify Help Center, automatic discounts · Purchase-gated referral: what it is and how the gate works · Referral and word of mouth

A gated program producing roughly 36 referred orders per cycle on a 1,000-customer base grows with the customer count, not with site traffic.

Modelled on stated assumptions: 1,000 customers, 30% ever share a link, 12% per-link conversion, one conversion per sharing customer. · Purchase-gated referral: what it is and how the gate works · Referral and word of mouth

9 ideas, ordered by setup effort, run from 30 minutes to 1 week, and the 2 cheapest use assets the brand already ships.

Derived from the effort ranking in this article. · Quirky ideas in affiliate marketing that still work · Affiliate marketing

A packaging insert carries a printed slug rather than a code, which recovers attribution from a channel that produces no click at all.

Derived from the vanity slug mechanic described for idea one. · Quirky ideas in affiliate marketing that still work · Affiliate marketing

Making the purchase the application converts a modelled 5.1% of customers against 1.1% through an application form.

Modelled on stated assumptions: an application funnel with completion and approval steps against a link that already exists. · Quirky ideas in affiliate marketing that still work · Affiliate marketing

A chain paying two levels at 6% and 3% costs 9% of order value, which fits under a 35% of contribution margin ceiling at 26% contribution margin.

Modelled on stated assumptions: the payout ceiling rule applied to a two-level chain. · Quirky ideas in affiliate marketing that still work · Affiliate marketing

Pairing a partner with a live delivery zone gives an exclusive offer with no code string that can be republished on an aggregator site.

Derived from the geo-gated offer mechanic described for idea nine. · Quirky ideas in affiliate marketing that still work · Affiliate marketing

Customers with three or more orders who have left a review activate at a modelled 34% when asked directly.

Modelled on stated assumptions: outreach acceptance and first-share rates by behavioural segment. · How to recruit your first 100 affiliates from your order list · Affiliate marketing

Follower-count-based outreach to 200 customers with large audiences models to 12 activations, against 68 from the top behavioural segment.

Modelled on stated assumptions: 6% activation on follower-selected outreach against 34% on the top behavioural segment. · How to recruit your first 100 affiliates from your order list · Affiliate marketing

The top two segments of a 5,000-customer base model to 153 activated affiliates, more than the hundred target.

Modelled on stated assumptions: 4% and 9% segment shares activating at 34% and 19%. · How to recruit your first 100 affiliates from your order list · Affiliate marketing

Discount-acquired single-order customers activate at a modelled 1.5%, the lowest of six segments.

Modelled on stated assumptions about what discount-acquired customers actually valued in the transaction. · How to recruit your first 100 affiliates from your order list · Affiliate marketing

Review activity separates three-order customers into 34% and 19% activation, the widest split from any single signal.

Modelled on stated assumptions comparing repeat customers with and without a review on file. · How to recruit your first 100 affiliates from your order list · Affiliate marketing

A delivery-triggered referral email models to 2.95% share contribution against 0.71% at purchase confirmation, roughly four times more.

Modelled on stated assumptions: 61% open and 11% click at delivery against 34% open and 6% click at purchase, with share rates of 44% and 35%. · Referral emails that get opened · Referral and word of mouth

Open rate rises from a modelled 34% at purchase confirmation to 61% at delivery confirmation, because delivery notifications are the most opened transactional email a store sends.

Modelled on stated assumptions about transactional email engagement by type. · Referral emails that get opened · Referral and word of mouth

Share rate among people who reach the referral page rises from 35% at purchase to 44% at delivery, a 9 point difference driven by having the product.

Modelled on stated assumptions: willingness to recommend rising once the product is in hand. · Referral emails that get opened · Referral and word of mouth

A send 30 days after delivery models to 0.64% share contribution, below the purchase confirmation it was meant to improve on.

Modelled on stated assumptions: 28% open, 6% click and 38% share as the moment recedes. · Referral emails that get opened · Referral and word of mouth

Two sends, at delivery and seven days after, model to 4.68% combined against 2.95% for delivery alone.

Derived by summing the modelled contributions of the two triggers. · Referral emails that get opened · Referral and word of mouth

Modelled redemption rises from 21% at a 24-hour window to 68% at 30 days, then only to 74% at 90 days.

Modelled on stated assumptions: redemption requiring a future order, concentrated in the first weeks after issuance. · Referral expiry windows: the setting nobody tunes · Referral and word of mouth

Extending from 30 to 90 days buys 6 percentage points of redemption while roughly tripling modelled outstanding liability.

Modelled on stated assumptions: liability accumulating with window length at a constant issuance rate. · Referral expiry windows: the setting nobody tunes · Referral and word of mouth

Median days to redeem runs 0.4 days at a 24-hour window and 26 days at 90 days, so the window largely sets the behaviour it measures.

Modelled on stated assumptions about urgency compressing or extending the decision. · Referral expiry windows: the setting nobody tunes · Referral and word of mouth

The redeeming order carries a modelled 9% higher average order value at a 30-day window, because a capped balance cannot be cleared on a minimum basket.

Modelled on stated assumptions: a 15% redemption cap against a ₹1,450 average order value. · Referral expiry windows: the setting nobody tunes · Referral and word of mouth

2 clocks are needed: an offer window measured in days and a coin expiry measured in months, because they control different things.

Derived from the distinction between a rotating live discount and an earned wallet balance. · Referral expiry windows: the setting nobody tunes · Referral and word of mouth

Purchase gating makes self-referral cost 1 full-price order every time, which converts an unlimited attack into one with a price attached.

Derived from the structural constraint described for pattern one. · Referral fraud patterns and how to design them out · Referral and word of mouth

Pattern 3, refund farming, is eliminated entirely by issuing coins only after the return window closes, at the cost of a delay customers must be told about upfront.

Derived from the structural constraint described for pattern three. · Referral fraud patterns and how to design them out · Referral and word of mouth

Pattern 2, ring referral, is removed by an acyclic chain rule where no participant may appear twice in their own upstream path, with no detection logic running.

Derived from the graph constraint described for pattern two. · Referral fraud patterns and how to design them out · Referral and word of mouth

Pattern 5, zone farming, is defended by a gate that reads Shopify's own delivery address rather than a cart attribute, which the shopper cannot edit.

Derived from the constraint described for pattern five, using the delivery address available inside a Shopify Function. · Referral fraud patterns and how to design them out · Referral and word of mouth

6 patterns, 6 constraints, and none of them requires a fraud score, a manual review queue, or a machine learning model.

Derived from the full pattern table in this article. · Referral fraud patterns and how to design them out · Referral and word of mouth

An order status page block contributes a modelled 2.9% share rate against 0.7% for a post-purchase email, roughly four times more.

Modelled on stated assumptions: 78% page view and 9% engagement against a 34% email open and 6% click, with share rates of 42% and 35%. · The referral link nobody clicks (and the one they do) · Referral and word of mouth

Making a share sheet the primary action rather than a copy button applies a modelled 1.4x multiplier across every placement.

Modelled on stated assumptions: removing the step where the sender has to decide what to write. · The referral link nobody clicks (and the one they do) · Referral and word of mouth

A cart placement contributes almost nothing to share rate, because the shopper has not yet had the experience they would be recommending.

Modelled on stated assumptions about timing relative to purchase and delivery. · The referral link nobody clicks (and the one they do) · Referral and word of mouth

Adding an account page balance contributes a further modelled 1.6 percentage points, but only where the reward is a visible running balance.

Modelled on stated assumptions: 24% account page visit rate, 22% engagement, 30% share rate. · The referral link nobody clicks (and the one they do) · Referral and word of mouth

The full placement stack models to about 7.3% participation against 0.7% for email alone.

Derived by summing the modelled placement contributions and applying the share channel multiplier. · The referral link nobody clicks (and the one they do) · Referral and word of mouth

A balance stored as a mutable field cannot answer what the liability was on a past date, which is the question an audit actually asks.

Derived from the schema comparison between a balance column and an append-only entry ledger. · Building a referral program your finance team will approve · Referral and word of mouth

At 60 referred orders a month and ₹203 of coins per order, issuance runs ₹12,180 monthly.

Modelled on stated assumptions: a three-level chain paying 14% of a ₹1,450 order value. · Building a referral program your finance team will approve · Referral and word of mouth

With 68% redemption and a 12-month expiry, modelled steady-state outstanding liability settles near ₹58,000, or about 4.8 months of issuance.

Modelled on stated assumptions: redemption concentrated in the first three months after issuance, remainder expiring at month 12. · Building a referral program your finance team will approve · Referral and word of mouth

Recognising the expense at confirmation rather than at issuance defers it by the length of the return window, typically 7 to 30 days.

Modelled on stated assumptions about pending coins becoming unconditional once the return window closes. · Building a referral program your finance team will approve · Referral and word of mouth

6 entry types cover every event a coin ledger produces: issue, confirm, reverse, redeem, expire, and adjust.

Derived from the ledger schema set out in this article. · Building a referral program your finance team will approve · Referral and word of mouth

14 checks cover a referral launch, and 9 of them are decisions rather than engineering.

Derived from the four-stage checklist in this article. · The referral program launch checklist · Referral and word of mouth

Contribution margin on a modelled ₹1,450 order is ₹569 after ₹83 of order costs, not the ₹652 gross margin figure most teams use.

Modelled on stated assumptions: 45% gross margin less payment fees, shipping subsidy and return losses. · The referral program launch checklist · Referral and word of mouth

A default chain of 8%, 4% and 2% costs ₹203 per order, which is 35.7% of that contribution margin.

Modelled on stated assumptions, applying the payout curve to the contribution margin figure above. · The referral program launch checklist · Referral and word of mouth

8 test scenarios should run before launch, of which partial refund and cap collision are the 2 that reliably surface bugs.

Derived from the pre-launch test list in this article. · The referral program launch checklist · Referral and word of mouth

Redemption cannot be meaningfully read before roughly week 6, because coins issued in week 2 require a second order to be spent.

Modelled on stated assumptions about repeat purchase intervals in a consumables category. · The referral program launch checklist · Referral and word of mouth

Script tags stop running on the order status page for non-Plus stores on 26 August 2026, having already stopped for Plus stores in August 2025.

Shopify developer documentation, ScriptTag resource (legacy) · The Shopify apps quietly slowing your checkout down · Shopify stack and extensions

Script tags now work only on vintage themes, leaving theme app extensions as the one supported route for any app in the Shopify App Store.

Shopify developer documentation, ScriptTag resource (legacy) · The Shopify apps quietly slowing your checkout down · Shopify stack and extensions

Shopify's web performance report runs on a rolling 28 day window, so an app installed yesterday will not show its full cost until next month.

Shopify Help Center, web performance reports · The Shopify apps quietly slowing your checkout down · Shopify stack and extensions

A 75th percentile LCP above 2,500ms is the line Shopify's own performance reporting treats as needing improvement.

Shopify performance blog, web performance report queries · The Shopify apps quietly slowing your checkout down · Shopify stack and extensions

A 120 kB app script loaded from a third-party origin models out to roughly 270ms of delay before the largest element paints.

Modelled on stated assumptions: 1ms of main-thread time per kB of JavaScript on a mid-range Android handset, plus 150ms to open a new third-party connection. · The Shopify apps quietly slowing your checkout down · Shopify stack and extensions

At 900 orders a month and ₹720 of margin per order, every 200ms of added delay costs about ₹6,480 if you assume a 1% conversion loss.

Modelled on stated assumptions: 900 orders, ₹1,800 AOV, 40% gross margin, 1% conversion loss per 200ms. · The Shopify apps quietly slowing your checkout down · Shopify stack and extensions

A Shopify store can have a maximum of 25 active automatic discounts, and app-based discounts count towards that total.

Shopify Help Center, combining discounts · The Shopify apps that quietly break each other · Shopify stack and extensions

When 2 product discounts target the same line item, Shopify applies only the better one instead of stacking both.

Shopify Help Center, combining discounts · The Shopify apps that quietly break each other · Shopify stack and extensions

A customer can apply a maximum of 5 product or order discount codes plus 1 shipping discount code to a single order.

Shopify Help Center, combining discounts · The Shopify apps that quietly break each other · Shopify stack and extensions

Automatic discounts created in the Shopify admin do not apply to the post-purchase page at checkout.

Shopify Help Center, automatic discounts · The Shopify apps that quietly break each other · Shopify stack and extensions

6 app categories produce 15 possible pairings, and 11 of them have a documented way to collide.

Derived from the conflict matrix on this page: six categories taken two at a time. · The Shopify apps that quietly break each other · Shopify stack and extensions

6 of those 15 pairings collide on the discount stack alone, which is where most support tickets originate.

Derived from the conflict matrix on this page. · The Shopify apps that quietly break each other · Shopify stack and extensions

An estimated 40% of Shopify stores render no app block anywhere in their theme.

Modelled on stated assumptions: six installed apps, 35% of apps shipping a placeable app block rather than an embed only, 40% merchant placement rate. · Your Shopify extensions nobody can ignore · Shopify stack and extensions

A default Online Store 2.0 theme exposes roughly 15 places where an app block can be added without editing theme code.

Approximate count of app-block-capable section slots across Dawn's home, product, collection and cart templates; the number varies by theme and version. · Your Shopify extensions nobody can ignore · Shopify stack and extensions

One well-placed app block is worth about Rs 1.08 lakh a month in incremental revenue at the modelled inputs.

Modelled on stated assumptions: 20,000 monthly sessions, a 0.3 percentage point conversion lift, Rs 1,800 average order value. · Your Shopify extensions nobody can ignore · Shopify stack and extensions

A Shopify Function adds 0 milliseconds of client-side JavaScript to your storefront.

Derived from Shopify's execution model: Functions run server side during cart and checkout evaluation and ship no browser code. · Your Shopify extensions nobody can ignore · Shopify stack and extensions

Of the 3 checkout surfaces an app can render on, only the in-checkout steps require Shopify Plus.

Shopify developer documentation, About checkout app extensions · Your Shopify extensions nobody can ignore · Shopify stack and extensions

A capped chained referral programme's worst-case discount per order stays at 10% however deep the chain runs.

Derived from capped-redemption arithmetic: the per-order cap binds total discount, so extra chain levels split the same pool rather than adding to it. · Your Shopify extensions nobody can ignore · Shopify stack and extensions

Shopify Scripts stopped executing entirely on 30 June 2026.

Shopify developer changelog, Scripts deprecation notice · Shopify Functions vs Scripts: what changed after Scripts died · Shopify stack and extensions

Editing and publishing Shopify Scripts was frozen on 15 April 2026, before execution stopped.

Shopify developer changelog, Scripts deprecation notice · Shopify Functions vs Scripts: what changed after Scripts died · Shopify stack and extensions

A Shopify Function must complete its logic inside 11 million WebAssembly instructions or the run fails.

Shopify Function APIs documentation, resource limits · Shopify Functions vs Scripts: what changed after Scripts died · Shopify stack and extensions

Function input is capped at 128 kB, which is why merchant configuration lives in metafields rather than in a payload.

Shopify developer changelog, Functions input size increase · Shopify Functions vs Scripts: what changed after Scripts died · Shopify stack and extensions

Only 1 route exists for a Function to reach an external service at runtime, and it is restricted to custom apps on Plus and Enterprise stores.

Shopify Discount Function API documentation, fetch target · Shopify Functions vs Scripts: what changed after Scripts died · Shopify stack and extensions

Daily rotation of an offer turns 900 potential checkout-time lookups into 30 metafield writes a month.

Derived: one metafield write per daily rotation across 30 days, against a 900 order month. · Shopify Functions vs Scripts: what changed after Scripts died · Shopify stack and extensions

A Function executes in under 5 milliseconds with no cold start, because it runs as compiled WebAssembly inside Shopify's platform rather than in a separately provisioned runtime.

Shopify Functions documentation · Shopify Functions vs Scripts: what changed after Scripts died · Shopify stack and extensions

Shopify Plus starts at $2,300 a month on a three-year term, or from $2,500 a month on a one-year term.

Shopify Plus pricing page · Shopify Plus features you're paying for and not using · Shopify stack and extensions

The Advanced plan is $399 a month billed monthly, which puts the Plus premium at $1,901 a month before anything else.

Shopify pricing page, US pricing verified August 2026 · Shopify Plus features you're paying for and not using · Shopify stack and extensions

Checkout UI extensions on all 3 pre-purchase steps, information, shipping and payment, are available only to stores on a Shopify Plus plan.

Shopify developer documentation, checkout UI extensions · Shopify Plus features you're paying for and not using · Shopify stack and extensions

Thank you and order status page extensions are available on every plan except Starter, so that surface is not a reason to be on Plus.

Shopify developer documentation, checkout app extensions · Shopify Plus features you're paying for and not using · Shopify stack and extensions

Splitting the $1,901 premium over eight exclusive capabilities puts each at $238 a month, so a store using three pays $1,188 a month for the five it does not.

Derived: $1,901 divided by eight Plus-exclusive capabilities. · Shopify Plus features you're paying for and not using · Shopify stack and extensions

Five unused Plus capabilities cost about ₹1,03,000 a month at ₹87 to the dollar.

Derived: $1,188 converted at a stated ₹87 exchange rate. · Shopify Plus features you're paying for and not using · Shopify stack and extensions

A ₹3,000 a month Shopify app breaks even at five incremental orders.

Modelled on stated assumptions: ₹1,800 AOV, 40% gross margin, ₹720 margin per order, ₹3,000 monthly fee. · The Shopify app stack that pays for itself in 30 days · Shopify stack and extensions

A referral app charging ₹4,000 a month breaks even when 0.8% of a 900 order month arrives through referral links.

Derived: ₹4,000 divided by ₹576 net margin per referred order, assuming ₹1,800 AOV, 40% gross margin, coin redemption capped at 8% of order value. · The Shopify app stack that pays for itself in 30 days · Shopify stack and extensions

Moving 10% of a media budget from a 2.0 ROAS channel to a 4.0 ROAS channel returns about six times a typical attribution fee in one month.

Modelled on stated assumptions: ₹5,00,000 monthly budget split evenly across two channels, 40% gross margin, ₹6,000 monthly attribution fee. · The Shopify app stack that pays for itself in 30 days · Shopify stack and extensions

An app that takes a fortnight to configure has 16 of its first 30 billed days left to produce a return.

Derived from a 30 day billing cycle and a 14 day configuration period. · The Shopify app stack that pays for itself in 30 days · Shopify stack and extensions

A modelled median stack of six apps costs about ₹7,400 a month, of which only two apps and ₹2,600 touch what a shopper actually decides.

Modelled on stated assumptions: a composite stack of email, reviews, shipping, support, analytics and page building at published entry-tier pricing. · Tools your Shopify store must have (and the three most stores skip) · Shopify stack and extensions

Adding the three skipped categories takes the revenue-touching share of a stack from 33% of apps to 56%.

Modelled on stated assumptions: six apps with two revenue-touching, rising to nine apps with five revenue-touching. · Tools your Shopify store must have (and the three most stores skip) · Shopify stack and extensions

A sitewide 10% discount costs about ₹1,813 of subsidy per incremental order against ₹725 for the same offer gated to one high-density zone.

Modelled on stated assumptions: 1,000 orders at ₹1,450, 8% lift sitewide, 12% of orders in the gated zone with a 20% lift there. · Tools your Shopify store must have (and the three most stores skip) · Shopify stack and extensions

A referral program that lives only in a post-purchase email models to 0.7% participation, from a 34% open rate, 6% click rate and 35% share rate.

Modelled on stated assumptions, multiplied through in the referral ledger section. · Tools your Shopify store must have (and the three most stores skip) · Shopify stack and extensions

If 14% of orders are touched by two paid channels, both channels claim them and your reported channel revenue exceeds actual revenue by that overlap.

Modelled on stated assumptions about last-click reporting in two ad platforms with no shared priority order. · Tools your Shopify store must have (and the three most stores skip) · Shopify stack and extensions

A single theme app extension can contain up to 30 app blocks, raised from 25 in February 2026.

Shopify developer changelog, app block limit increase · Theme app extensions: the free real estate on your store · Shopify stack and extensions

Theme app extension blocks render on 0 checkout pages, and that includes the order status page.

Shopify developer documentation, configure theme app extensions · Theme app extensions: the free real estate on your store · Shopify stack and extensions

App embed blocks can load scripts on specific pages only, which a script tag cannot do.

Shopify developer documentation, configure theme app extensions · Theme app extensions: the free real estate on your store · Shopify stack and extensions

A referral block on the product page gets 35 times the impressions of the post-purchase surface and produces nothing, because none of those visitors have bought yet.

Derived: 31,500 modelled product page views against 900 orders a month on stated assumptions. · Theme app extensions: the free real estate on your store · Shopify stack and extensions

Ten percent of 900 buyers sharing a link, at 15% conversion, models to 13.5 referred orders a month.

Modelled on stated assumptions: 900 orders, 10% share rate on post-purchase surfaces, 15% of shared links converting. · Theme app extensions: the free real estate on your store · Shopify stack and extensions

A modelled typical affiliate stack covers 2.4 of the four jobs, leaving the remainder to manual work nobody scheduled.

Modelled on stated assumptions: coverage scored across the tool categories in the matrix in this article. · Tools to grow sales through affiliate marketing · Affiliate marketing

Payout is covered fully by 3 of the 6 tool categories and left partial, manual or absent by the other 3, because it needs verification, tax handling and money movement rather than a dashboard.

Derived from the coverage matrix, where payout is fully covered by only two of six categories. · Tools to grow sales through affiliate marketing · Affiliate marketing

Fraud control is structural rather than a feature, and just 1 of the 6 tool categories covers it fully: the one where every affiliate had to buy first.

Derived from the coverage matrix and the cost-to-fake logic described in the fraud section. · Tools to grow sales through affiliate marketing · Affiliate marketing

An attribution platform scores full on 1 of the 4 jobs, tracking, and none at all on recruitment and payout, which is why it is a complement rather than an affiliate tool.

Derived from the coverage matrix in this article. · Tools to grow sales through affiliate marketing · Affiliate marketing

Recruitment through an application form converts at a modelled 1.1% of customers shown the ask, against 5.1% when the purchase is the application.

Modelled on stated assumptions: an application funnel with completion and approval steps against a purchase-gated link that already exists. · Tools to grow sales through affiliate marketing · Affiliate marketing

₹10 lakh a month at a ₹1,800 average order value is 556 orders, or about 18 orders a day.

Derived: ₹10,00,000 divided by ₹1,800, spread across 30 days. · What to install first when you cross ₹10L a month · Shopify stack and extensions

Attribution starts paying back at roughly ₹75,000 of monthly media spend, which is about ₹3 lakh of revenue at a 25% spend ratio.

Modelled on stated assumptions: 10% of budget moved from a 2.0 ROAS channel to a 4.0 ROAS channel, 40% gross margin, ₹6,000 monthly fee. · What to install first when you cross ₹10L a month · Shopify stack and extensions

A referral program's threshold is a stock rather than a flow: about 2,300 lifetime buyers, not a monthly revenue figure.

Modelled: 2% of lifetime buyers share in a month, 15% of shared links convert, against a 6.9 order monthly break-even. · What to install first when you cross ₹10L a month · Shopify stack and extensions

One sitewide 10% promotion across a ₹10 lakh month gives away ₹1,00,000, which is roughly 40 times a typical offer app's monthly fee.

Derived: 10% of ₹10,00,000 against an illustrative ₹2,500 monthly app fee. · What to install first when you cross ₹10L a month · Shopify stack and extensions

Helpdesk software arrives at about ₹12 lakh a month, on 15 tickets per 100 orders and a 100 ticket threshold.

Derived: 100 tickets divided by 0.15 tickets per order is 667 orders, at ₹1,800 AOV. · What to install first when you cross ₹10L a month · Shopify stack and extensions

₹10 lakh of jewellery at ₹15,000 AOV is 67 orders a month, which fails every order-count threshold on this page.

Derived: ₹10,00,000 divided by ₹15,000. · What to install first when you cross ₹10L a month · Shopify stack and extensions

A click costs effectively nothing to fabricate, which makes it the only 1 of the 4 candidate triggers with unbounded fraud exposure.

Derived from the cost-to-fake comparison across the four trigger events in this article. · What to reward: the click, the signup, or the second order · Referral and word of mouth

A signup trigger costs an attacker 1 disposable email address per reward, which is a price near zero at any scale.

Derived from the same comparison, assuming no payment step between signup and reward. · What to reward: the click, the signup, or the second order · Referral and word of mouth

A first settled order costs an attacker a full-price purchase, which exceeds the reward at any payout under 100% of order value.

Modelled on stated assumptions: a 14% chain payout against a full-price order with a real product shipped. · What to reward: the click, the signup, or the second order · Referral and word of mouth

Triggering on the second order pushes reward issuance a modelled 34 days later on a category with a 34-day repeat interval.

Modelled on stated assumptions about median repeat purchase timing in a consumables category. · What to reward: the click, the signup, or the second order · Referral and word of mouth

A 2-stage reward, issuing at the first order and confirming after the return window closes, keeps the fraud exposure of the order trigger while removing refund farming.

Derived from the pending-issuance design described in this article. · What to reward: the click, the signup, or the second order · Referral and word of mouth

Customers with three or more orders who have left a review are 4% of the base and refer at a modelled 31%, over five times the 6% average.

Modelled on stated assumptions: six behavioural segments scored on referral likelihood across a full customer base. · Who actually refers, and how to find them before they do · Referral and word of mouth

Discount-acquired single-order customers are 28% of the base and refer at a modelled 1.6%, roughly a quarter of the average rate.

Modelled on stated assumptions, from the same six-segment breakdown. · Who actually refers, and how to find them before they do · Referral and word of mouth

Review activity is the strongest single predictor, separating three-order customers into 31% and 14% referral likelihood.

Modelled on stated assumptions comparing repeat customers with and without a review on file. · Who actually refers, and how to find them before they do · Referral and word of mouth

Basket breadth doubles referral likelihood among two-order customers, from 6% for narrow baskets to 12% for three or more categories.

Modelled on stated assumptions about category spread as a proxy for engagement with the range. · Who actually refers, and how to find them before they do · Referral and word of mouth

The top two segments are 13% of the customer base and produce a modelled 42% of referrals.

Derived from the segment shares and likelihoods in the referral likelihood table. · Who actually refers, and how to find them before they do · Referral and word of mouth

A halving payout curve of 8%, 4%, 2%, 1% and 0.5% totals 15.5% of order value at depth 5, against 25% for a flat 5% curve.

Modelled on stated assumptions, summed cumulatively across depths 1 to 5. · How deep should a referral chain go? A margin-first answer · Referral and word of mouth

A 14% payout consumes 35.7% of contribution margin on a brand whose contribution margin is 39% of order value.

Modelled on stated assumptions: ₹1,450 AOV, 45% gross margin, ₹83 of order costs giving ₹569 contribution margin. · How deep should a referral chain go? A margin-first answer · Referral and word of mouth

The same 14% payout consumes 70% of contribution margin on a brand at 20% contribution margin, which no incrementality rate rescues.

Modelled on stated assumptions, applying the same payout to a thinner margin structure. · How deep should a referral chain go? A margin-first answer · Referral and word of mouth

At a 35% ceiling, a brand with 25% contribution margin can afford 8.75% of order value in total payout, which is depth 2 on a halving curve.

Derived from the ceiling rule applied to the modelled margin bands. · How deep should a referral chain go? A margin-first answer · Referral and word of mouth

A halving curve converges, so depth 5 costs only 1.5 percentage points more than depth 3, while a flat curve grows without limit.

Derived from the cumulative payout table for the three curve shapes. · How deep should a referral chain go? A margin-first answer · Referral and word of mouth

Four numbers reviewed weekly cover the decisions a brand under 1,000 orders a month actually makes.

Derived from the weekly template in this article. · Attribution for a brand with nobody to run it · Attribution and measurement

Blended acquisition cost, total spend divided by new customers, avoids the modelled 26% overcount produced by summing platform-reported conversions.

Modelled on stated assumptions: four platforms claiming 1,260 conversions against 1,000 actual orders. · Attribution for a brand with nobody to run it · Attribution and measurement

An unattributed share that moves more than 5 points week over week usually indicates a tracking regression rather than a change in demand.

Modelled on stated assumptions about the stability of dark-channel demand over short periods. · Attribution for a brand with nobody to run it · Attribution and measurement

The review takes 15 minutes because 3 of the 4 numbers come from Shopify's own reports without any additional tooling.

Derived from the data sources listed for each number in the template. · Attribution for a brand with nobody to run it · Attribution and measurement

New customer share is the earliest warning of an acquisition problem, moving weeks before blended acquisition cost does.

Modelled on stated assumptions: repeat orders sustaining revenue while new customer volume declines. · Attribution for a brand with nobody to run it · Attribution and measurement

7 isolation checks separate a genuinely multi-tenant tool from a single-tenant one with a client filter on the report.

Derived from the tenant-isolation test set out in this article. · Attribution for agencies running ten brands at once · Attribution and measurement

A shared slug namespace across 10 clients makes a collision likely once any 2 brands work with the same creator.

Modelled on stated assumptions: overlapping creator rosters within a category, with slugs derived from creator names. · Attribution for agencies running ten brands at once · Attribution and measurement

One agency-wide cookie domain leaks a customer's presence across all 10 client brands, since the same browser carries identifiers set by every brand.

Modelled on stated assumptions about cookies scoped to an agency domain rather than to each client's own domain. · Attribution for agencies running ten brands at once · Attribution and measurement

Portfolio benchmarking requires aggregate-only exposure: a category median across 10 clients reveals nothing, while a per-client column reveals everything.

Derived from the aggregation rules described in the portfolio section. · Attribution for agencies running ten brands at once · Attribution and measurement

Client offboarding should delete tenant data on a stated schedule, and a tool with no deletion path leaves former clients' data inside your reporting indefinitely.

Derived from the deletion check in the isolation test. · Attribution for agencies running ten brands at once · Attribution and measurement

Eleven apps at ₹31,000 a month have to produce 43 incremental orders between them just to break even.

Derived: ₹31,000 divided by ₹720 of gross margin per order at ₹1,800 AOV and 40% margin. · How to audit your Shopify app stack in an afternoon · Shopify stack and extensions

Cancelling the four lowest-scoring apps on the sample scorecard returns ₹98,400 over a year.

Derived: ₹8,200 a month across four apps, over twelve months. · How to audit your Shopify app stack in an afternoon · Shopify stack and extensions

Shopify's web performance report runs on a rolling 28 day window, so a cancellation takes a month to read cleanly.

Shopify Help Center, web performance reports · How to audit your Shopify app stack in an afternoon · Shopify stack and extensions

Script tags stop running on the order status page for non-Plus stores on 26 August 2026.

Shopify developer documentation, ScriptTag resource (legacy) · How to audit your Shopify app stack in an afternoon · Shopify stack and extensions

Testing 3 cancellations properly, 1 at a time with 14 days between each, takes 6 weeks.

Derived: three cancellations at 14 days apart. · How to audit your Shopify app stack in an afternoon · Shopify stack and extensions

A referral cohort reaches cumulative contribution margin of ₹652 in month zero against an acquisition cost of ₹203, paying back immediately.

Modelled on stated assumptions: ₹1,450 AOV, 45% gross margin, chained coin payout of 14% of order value. · Cohort reporting your CFO will read · Attribution and measurement

A paid social cohort reaches ₹1,031 of cumulative contribution margin by month five against a ₹1,091 acquisition cost, still short of payback.

Modelled on stated assumptions: monthly repeat contribution decaying from ₹105 in month one to ₹55 in month five. · Cohort reporting your CFO will read · Attribution and measurement

Referral cohorts contribute ₹196 in month one against ₹105 for paid social, an 87% difference in early repeat value.

Modelled on stated assumptions about repeat rates by acquisition channel across the same six-month window. · Cohort reporting your CFO will read · Attribution and measurement

The one-page layout carries four columns of metadata and six months of cells, which is 10 numbers per cohort row.

Derived from the template layout described in this article. · Cohort reporting your CFO will read · Attribution and measurement

Reporting gross margin instead of contribution margin overstates modelled cumulative value by ₹83 per order.

Modelled on stated assumptions: payment fees, shipping subsidy and return losses excluded from gross margin but real in cash terms. · Cohort reporting your CFO will read · Attribution and measurement

At 120 orders a month per arm, detecting a 15% lift requires roughly three months, because one month's natural variation is close to the effect size.

Modelled on stated assumptions: order counts varying with the square root of volume, and a difference of about 2.8 standard errors treated as readable. · How to measure a channel that has no clicks · Attribution and measurement

A vanity slug spoken aloud in a podcast recovers a modelled 23% of that channel's driven orders, against 0% from a pixel.

Modelled on stated assumptions: listeners who both remember the URL and type it, on a short, sayable path. · How to measure a channel that has no clicks · Attribution and measurement

Word of mouth takes 21% of orders in a modelled post-purchase survey and appears nowhere in click data.

Modelled on stated assumptions: 1,000 orders scored by survey and by pixel in the same month. · How to measure a channel that has no clicks · Attribution and measurement

Matching holdout zones on baseline order volume, category mix and delivery time reduces modelled pre-period divergence to under 4%.

Modelled on stated assumptions about zone selection using three months of pre-period order data. · How to measure a channel that has no clicks · Attribution and measurement

A geo-holdout costs real money by design: withholding a channel from 120 orders a month forgoes a modelled 18 orders if the channel works.

Derived from the modelled 15% lift applied to the held-back arm. · How to measure a channel that has no clicks · Attribution and measurement

A client-side purchase pixel captures a modelled 934 of 1,000 orders, a 6.6% loss before attribution logic runs.

Modelled on stated assumptions: ad blockers, tracking prevention, sessions ending early and in-app browsers. · Server-side tracking for stores that can't afford a data team · Attribution and measurement

Ad and script blockers account for a modelled 3.1 of those 6.6 percentage points, the single largest component.

Modelled on stated assumptions about blocker adoption on mobile browsers in a D2C audience. · Server-side tracking for stores that can't afford a data team · Attribution and measurement

An orders/create webhook captures a modelled 998 of 1,000 orders, with the remainder recovered by Shopify's retry schedule.

Modelled on stated assumptions: transient endpoint failures resolved within the retry window. · Server-side tracking for stores that can't afford a data team · Attribution and measurement

Passing the same event_id from both the pixel and the server prevents double counting on a modelled 934 orders seen twice.

Modelled on stated assumptions: every pixel-captured order also arriving through the webhook path. · Server-side tracking for stores that can't afford a data team · Attribution and measurement

The full setup is 3 components: 1 webhook subscription, 1 attribution table, and 1 deduplication key.

Derived from the implementation steps in this article. · Server-side tracking for stores that can't afford a data team · Attribution and measurement

Reported CAC of ₹1,091 rises to a fully loaded ₹1,364 once agency fees, first-order discount and referral payout are included, a 25% understatement.

Modelled on stated assumptions: ₹6,00,000 ad spend, 550 incremental new customers, ₹90,000 of agency and creative cost. · What your CAC number is actually missing · Attribution and measurement

Agency and creative fees add ₹164 per new customer, the single largest missing line in the modelled calculation.

Modelled on stated assumptions: ₹90,000 monthly agency and production cost across 550 new customers. · What your CAC number is actually missing · Attribution and measurement

First-order discount adds ₹87 per new customer when 60% of acquisition orders carry an average 10% discount.

Modelled on stated assumptions: ₹1,450 AOV with discounting concentrated on first purchases. · What your CAC number is actually missing · Attribution and measurement

Payment fees, shipping subsidy and return losses total ₹83 per order and belong in contribution margin, which falls from ₹652 to ₹569.

Modelled on stated assumptions: 2% gateway fee, ₹70 shipping on 62% of orders, and returns at a modelled 5.9% of orders. · What your CAC number is actually missing · Attribution and measurement

Fully loaded CAC against corrected contribution margin means payback needs 2.4 orders per customer rather than 1.7.

Derived from ₹1,364 fully loaded CAC divided by ₹569 of corrected contribution margin. · What your CAC number is actually missing · Attribution and measurement

5 apps produce 10 possible pairings and 1 app produces none, so failure surface falls faster than app count does.

Derived: n(n-1)/2 pairings across n apps. · Replacing five Shopify apps with one · Shopify stack and extensions

Cutting a stack from eleven apps to seven removes 34 of 55 pairings, a 62% reduction from a 36% cut in app count.

Derived: 55 pairings across eleven apps against 21 across seven. · Replacing five Shopify apps with one · Shopify stack and extensions

Five rewards apps at ₹12,500 a month replaced by one at ₹4,000 returns ₹1,02,000 over a year.

Derived: ₹8,500 monthly saving across twelve months. · Replacing five Shopify apps with one · Shopify stack and extensions

Five apps shipping 360 kB of JavaScript across four third-party origins model to between 500ms and 960ms of added delay.

Modelled on stated assumptions: 1ms of main-thread time per kB, 150ms per new third-party origin, upper bound assuming no overlap. · Replacing five Shopify apps with one · Shopify stack and extensions

Removing 5 apps properly, 1 every 14 days, takes 10 weeks.

Derived: five cancellations at 14 days apart. · Replacing five Shopify apps with one · Shopify stack and extensions

Four ad platforms claim 1,260 conversions in a month where Shopify recorded 1,000 orders, a 26% overcount before any analysis.

Modelled on stated assumptions: Meta, Google Ads, email and TikTok reporting at their default windows with no cross-platform deduplication. · Your attribution is wrong. Here's how wrong. · Attribution and measurement

Of those 1,260 claimed conversions, a modelled 550 are incremental, or 44% of the claim.

Modelled on stated assumptions about view-through inflation, brand search cannibalisation and email claiming returning customers. · Your attribution is wrong. Here's how wrong. · Attribution and measurement

Roughly 450 orders a month, 45% of the total, are claimed by no ad platform at all.

Modelled on stated assumptions: the residual after paid claims are deduplicated against a 1,000-order base. · Your attribution is wrong. Here's how wrong. · Attribution and measurement

At ₹6,00,000 of monthly spend, the claimed conversion count implies a CAC of ₹476 against a modelled ₹1,091.

Modelled on stated assumptions: total spend divided by claimed conversions versus by modelled incremental orders. · Your attribution is wrong. Here's how wrong. · Attribution and measurement

A modelled 210 of Meta's 640 claimed conversions are view-through, meaning nobody clicked anything.

Modelled on stated assumptions using a 7-day click and 1-day view attribution setting. · Your attribution is wrong. Here's how wrong. · Attribution and measurement

A sitewide 10% discount models to a net margin loss of ₹104,400 a month on 1,000 orders, the worst result of the seven mechanics tested.

Modelled on stated assumptions: 1,000 baseline orders, ₹1,450 AOV, 45% gross margin, 8% incremental volume lift. · Every Shopify discount type, ranked by margin damage · Shopify stack and extensions

Gating the same 10% offer to one high-density pincode zone costs ₹725 of subsidy per incremental order against ₹1,813 sitewide.

Modelled on stated assumptions: 12% of orders inside the gated zone, 20% lift there, against an 8% lift applied to the whole book sitewide. · Every Shopify discount type, ranked by margin damage · Shopify stack and extensions

A free gift with a ₹400 retail price and ₹180 COGS is a 27% nominal discount that costs 12% of order value.

Modelled on stated assumptions: ₹1,450 AOV, gift costed at COGS rather than retail because that is what the P&L sees. · Every Shopify discount type, ranked by margin damage · Shopify stack and extensions

A cart-value threshold that moves 12% of orders up by ₹550 returns ₹4,500 a month from top-ups, against ₹37,800 given to orders already above the line.

Modelled on stated assumptions: threshold at ₹1,999, 18% of orders already above it, 10% off on a ₹2,100 qualifying order. · Every Shopify discount type, ranked by margin damage · Shopify stack and extensions

A three-level coin chain paying 8%, 4% and 2% costs ₹203 per acquired order against ₹652 of gross margin on that order.

Modelled on stated assumptions: ₹1,450 AOV, 45% gross margin, coins issued only after a referred purchase settles. · Every Shopify discount type, ranked by margin damage · Shopify stack and extensions

Scored on the same 100 orders, influencer takes 24 orders under first-touch and 6 under last-touch, an 18-order swing.

Modelled on stated assumptions: a D2C store with a 9-day median time to purchase and six active channels. · First-touch vs last-touch for D2C: pick one and defend it · Attribution and measurement

Email moves the other way, from 5 orders under first-touch to 21 under last-touch.

Modelled on stated assumptions, from the same 100-order sample scored under both models. · First-touch vs last-touch for D2C: pick one and defend it · Attribution and measurement

Averaging the two models gives influencer 15 orders, a number that describes no real event in the customer journey.

Derived from the modelled first-touch and last-touch scores for the same channel. · First-touch vs last-touch for D2C: pick one and defend it · Attribution and measurement

Under last-touch, direct rises from 8 orders to 19, absorbing demand created by channels that no longer get credit.

Modelled on stated assumptions about returning shoppers typing the brand name rather than clicking a tracked link. · First-touch vs last-touch for D2C: pick one and defend it · Attribution and measurement

Brand paid search takes 3 orders under first-touch and 9 under last-touch, the clearest case of a channel closing sales it did not create.

Modelled on stated assumptions, from the same 100-order sample. · First-touch vs last-touch for D2C: pick one and defend it · Attribution and measurement

Code-only tracking captures a modelled 34% of sales genuinely driven by a creator, because most of the audience never types the code.

Modelled on stated assumptions: 100 creator-driven orders, code entry limited to viewers who both remember and are still in the checkout flow. · Tracking influencer sales without a discount code · Attribution and measurement

A unique redirect slug with a 30-day first-touch cookie raises modelled coverage to 61%.

Modelled on stated assumptions: click-through tracked at the redirect, cookie surviving to purchase in most but not all sessions. · Tracking influencer sales without a discount code · Attribution and measurement

Adding a post-purchase confirmation question as a third signal takes modelled coverage to 78%.

Modelled on stated assumptions: survey response rate on the order status page and a share of respondents naming the creator. · Tracking influencer sales without a discount code · Attribution and measurement

A modelled 22% of creator-driven sales remain unattributable to any signal, and should be reported as unattributed rather than reallocated.

Derived from the modelled coverage of the three combined signals against the 100-order base. · Tracking influencer sales without a discount code · Attribution and measurement

Codes leak: a modelled 19% of code redemptions come from coupon aggregator sites rather than the creator's audience.

Modelled on stated assumptions about public codes being indexed and republished within days of a campaign going live. · Tracking influencer sales without a discount code · Attribution and measurement

A referral ask sent only by post-purchase email models to 0.7% participation, from a 34% open rate, 6% click rate and 35% share rate.

Modelled on stated assumptions, multiplied through in the placement section. · Why your referral program stalled at 3% and how to restart it · Referral and word of mouth

Adding one order status page block models participation to about 3.6%, which is where most stalled programs are sitting.

Modelled on stated assumptions: 78% of buyers view the order status page, 9% engage with the block, 42% of those share. · Why your referral program stalled at 3% and how to restart it · Referral and word of mouth

A visible coin balance on the account page models a further 1.6 percentage points of participation on top of email and order status page placement.

Modelled on stated assumptions: 24% account page visit rate, 22% engagement, 30% share rate. · Why your referral program stalled at 3% and how to restart it · Referral and word of mouth

A reward stated as a running coin balance models 1.6x the share rate of the same value stated as a percentage off a future order.

Modelled on stated assumptions, with the reward legibility multipliers listed by reward type. · Why your referral program stalled at 3% and how to restart it · Referral and word of mouth

Without a visible balance or a chain, modelled repeat-share rate after the first referral is 18%, against 44% with both.

Modelled on stated assumptions about the second-order gap, working shown in that section. · Why your referral program stalled at 3% and how to restart it · Referral and word of mouth

Modelled 80th percentile time to purchase is 4 days for consumables, 14 days for apparel, and 47 days for considered high-ticket categories.

Modelled on stated assumptions: three synthetic D2C time-to-purchase distributions built from typical repeat, seasonal and research-led buying patterns. · The attribution window that matches your buying cycle · Attribution and measurement

A default 7-day window captures a modelled 92% of consumable orders but only 34% of high-ticket ones.

Modelled on stated assumptions, read off the same three percentile distributions. · The attribution window that matches your buying cycle · Attribution and measurement

Running a 7-day window on a high-ticket category discards a modelled 66% of the journey before any attribution model executes.

Derived from the modelled high-ticket distribution, where the median order arrives 16 days after first session. · The attribution window that matches your buying cycle · Attribution and measurement

Extending a consumables window from 4 days to 30 adds a modelled 6 percentage points of coverage and admits a large share of already-loyal repeat buyers.

Modelled on stated assumptions about repeat purchase intervals overlapping a long lookback. · The attribution window that matches your buying cycle · Attribution and measurement

First-time and repeat customers differ by a modelled 11 days at the 80th percentile in the apparel distribution, which argues for two windows rather than one.

Modelled on stated assumptions: repeat buyers converting faster because the consideration already happened. · The attribution window that matches your buying cycle · Attribution and measurement

Word of mouth takes 21% of orders in a modelled post-purchase survey and 0% in pixel data, the single largest disagreement between the two.

Modelled on stated assumptions: 1,000 orders, survey and pixel shares scored on the same month. · Post-purchase surveys vs pixel data: use both, weight one · Attribution and measurement

Email takes 19% of orders by pixel and 6% by survey, because customers rarely name a channel that only timed their purchase.

Modelled on stated assumptions, from the same 1,000-order sample. · Post-purchase surveys vs pixel data: use both, weight one · Attribution and measurement

Creator and influencer traffic takes 6% by pixel and 17% by survey, a gap of 11 percentage points.

Modelled on stated assumptions about creator discovery happening without a trackable click. · Post-purchase surveys vs pixel data: use both, weight one · Attribution and measurement

A modelled 34% survey response rate means every survey percentage is measured on roughly a third of orders and scaled to the rest.

Modelled on stated assumptions: one question on the order status page with no incentive attached. · Post-purchase surveys vs pixel data: use both, weight one · Attribution and measurement

Blending with a stated rule moves a modelled 11% of orders between channels compared with using pixel data alone.

Derived from the modelled channel-level gaps and the weighting rule applied in the blending section. · Post-purchase surveys vs pixel data: use both, weight one · Attribution and measurement

A five-step resolution order assigns a modelled 812 of 1,000 orders to exactly one channel, leaving 188 unattributed.

Modelled on stated assumptions: code, cookie and publication signals applied in fixed priority against a 1,000-order month. · Building a single source of truth for marketing spend · Attribution and measurement

Step three, last-touch cookie to influencer, resolves 143 orders, the largest single step in the modelled order.

Modelled on stated assumptions about creator-driven sessions that convert without a code being entered. · Building a single source of truth for marketing spend · Attribution and measurement

Without a stated priority, a modelled 260 orders are claimed by two or more channels and appear twice in reporting.

Modelled on stated assumptions: four platforms claiming 1,260 conversions against 1,000 actual orders. · Building a single source of truth for marketing spend · Attribution and measurement

Changing the priority order mid-quarter invalidates trend comparison for a modelled 9% of orders, the share that resolves differently under the two rules.

Modelled on stated assumptions about orders carrying more than one competing signal. · Building a single source of truth for marketing spend · Attribution and measurement

Referral orders arriving through a coin link resolve at step three or earlier, moving a modelled 60 orders a month out of the direct bucket.

Modelled on stated assumptions: 60 referred orders on a 1,000-order base, each carrying its own link identifier. · Building a single source of truth for marketing spend · Attribution and measurement

A modelled unmanaged account carries 47 distinct utm_source values representing 9 real channels.

Modelled on stated assumptions: multiple people tagging links by hand over eighteen months with no naming table. · UTM hygiene: the thirty-minute fix worth lakhs · Attribution and measurement

Case variants alone account for 21 of those 47 values, since analytics platforms treat Facebook and facebook as different sources.

Modelled on stated assumptions about mixed manual entry across desktop and mobile composition. · UTM hygiene: the thirty-minute fix worth lakhs · Attribution and measurement

Collapsing 47 values to 9 moves a modelled 34% of sessions out of rows too small to appear in a default report view.

Modelled on stated assumptions: long-tail spellings each holding under 1% of sessions and falling below reporting thresholds. · UTM hygiene: the thirty-minute fix worth lakhs · Attribution and measurement

One rule, lowercase everything, resolves a modelled 80% of the mess without touching taxonomy at all.

Modelled on stated assumptions: 21 of 26 excess values in the sample differ only by capitalisation or trailing whitespace. · UTM hygiene: the thirty-minute fix worth lakhs · Attribution and measurement

A three-field taxonomy with fixed vocabularies produces 9 sources and 24 valid combinations, small enough to hold in a single sheet.

Modelled on stated assumptions: 9 sources, a controlled medium list, and campaign names generated rather than typed. · UTM hygiene: the thirty-minute fix worth lakhs · Attribution and measurement

On a modelled 1,000-order month, GA4 records 934 purchase events, a 6.6% client-side loss before any attribution logic runs.

Modelled on stated assumptions: ad blockers, tracking prevention and sessions terminated before the purchase event fires. · What to do when Shopify, Meta, and GA4 disagree · Attribution and measurement

A modelled 210 of Meta's 640 reported conversions are view-through, recorded with no click at all.

Modelled on stated assumptions using a 7-day click, 1-day view attribution setting. · What to do when Shopify, Meta, and GA4 disagree · Attribution and measurement

Restating Meta's conversions from ad-interaction date to order date moves a modelled 4.2% of them into a different calendar month.

Modelled on stated assumptions: a 9-day median time to purchase against month boundaries. · What to do when Shopify, Meta, and GA4 disagree · Attribution and measurement

After reconciliation, the three systems agree on a modelled 812 of 1,000 orders, leaving 188 that only one system can see.

Modelled on stated assumptions, worked through in the reconciliation section. · What to do when Shopify, Meta, and GA4 disagree · Attribution and measurement

Averaging the three reported totals gives 858 orders, a number 14% below the actual 1,000 and matching no system's definition.

Derived from the modelled reported totals of 1,000, 934 and 640. · What to do when Shopify, Meta, and GA4 disagree · Attribution and measurement