Attribution and measurement
15 of 14 planned articles published.
Attribution for a brand with nobody to run it A brand with no analyst needs four numbers reviewed weekly: total orders, new customer share, blended acquisition cost, and unattributed share. Fifteen minutes, one sheet, no dashboard. Everything more sophisticated goes unread and therefore changes nothing. Attribution for agencies running ten brands at once Agency attribution needs per-client data isolation at the storage layer, not at the report layer. Run a tenant-isolation test before buying any tool: namespace collision, cookie scope, export boundaries, role separation, token scope, deletion, and portfolio aggregation. Cohort reporting your CFO will read A cohort report a finance team will read fits on one page: acquisition month and channel down the side, months since acquisition across the top, cumulative contribution margin per customer in the cells, and the payback month marked. Everything else belongs in an appendix. How to measure a channel that has no clicks Channels with no click surface are measured by geo-holdout rather than by tracking. Split matched delivery zones into exposed and held-back groups, compare total orders rather than attributed ones, and run long enough that the effect exceeds the natural variation in order counts. Server-side tracking for stores that can't afford a data team Webhook-first attribution treats the Shopify order as the event rather than a browser pixel, which removes the 6.6% of orders a client-side pixel loses to blockers and broken sessions. It needs one webhook subscription, one storage table, and a deduplication key. What your CAC number is actually missing Most quoted CAC figures contain ad spend and nothing else. Adding agency and creative fees, first-order discount, and referral payout takes a modelled ₹1,091 to ₹1,364. Payment fees, shipping subsidy and return losses belong in contribution margin, not CAC, and are frequently in neither. First-touch vs last-touch for D2C: pick one and defend it First-touch and last-touch attribution answer different questions and should be reported as two separate columns, never averaged. Scored on the same 100 orders, first-touch credits discovery channels like influencer and paid social, while last-touch credits closing channels like email and direct. Tracking influencer sales without a discount code Discount codes capture only the fraction of an influencer's audience that remembers to type them. Replacing the code with a unique redirect slug, a first-touch cookie and a post-purchase confirmation question takes modelled coverage of genuinely creator-driven sales from 34% to 78%. The attribution window that matches your buying cycle Attribution windows should be derived from the 80th percentile of your own time from first session to order, not taken from a platform default. Modelled across three D2C categories that gives 4 days for consumables, 14 for apparel, and 47 for considered high-ticket purchases. Post-purchase surveys vs pixel data: use both, weight one Pixels measure precisely what they can see and are blind to every channel without a click. Surveys see the dark channels and remember imprecisely. Weight pixel data for channels with a click surface, survey data for channels without, and state the rule on every report. Building a single source of truth for marketing spend A single source of truth is one resolver that assigns each order to exactly one channel through a fixed priority order, with every dashboard built as a report on top of it. The priority order matters less than writing it down and never changing it mid-quarter. What to do when Shopify, Meta, and GA4 disagree Shopify, Meta and GA4 disagree because they count different events, timestamp them differently, and apply different attribution windows. Reconcile by fixing Shopify orders as the denominator, restating everything on order date, separating view-through, then reporting the residual instead of averaging. Attribution window: what it is and how window length changes credit An attribution window is the period after a click or impression during which a resulting sale is still credited to that touchpoint. A 7-day window credits purchases made within seven days; anything later is credited elsewhere. Your attribution is wrong. Here's how wrong. Ad platforms typically claim more conversions than a store has orders, because each counts a different event inside a different window with no shared deduplication. On a modelled 1,000-order month, four platforms claim 1,260 conversions while roughly 45% of orders are claimed by nobody at all. UTM hygiene: the thirty-minute fix worth lakhs Unmanaged UTM tagging produces dozens of spellings for a handful of real channels, splitting every channel's performance across rows nobody adds back together. Lowercasing everything and generating tags from a fixed table instead of typing them fixes most of it in half an hour.