# Chained referrals: paying three people for one sale, profitably

> Chained referral programs stay profitable at three levels paying roughly 8%, 4% and 2% of order value, or 14% total against a 45% gross margin. Marginal return turns negative at level five, where the payout buys almost no additional sharing behaviour.

**Source:** https://blog.korant.online/chained-referrals-paying-three-people
**Published:** 2026-08-20
**Author:** Nayak - Builds checkout, referral and attribution tooling for Shopify D2C brands

## Key facts

- 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.)
- 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.)
- 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.)
- 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.)
- 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%.)

The objection arrives in the first meeting, every time, usually from whoever owns the P&L. Paying one person for a sale is a discount. Paying three people for the same sale sounds like paying three times.

It is not, and the reason is that the three payouts are not three prices for the same thing. Level one buys the share. Levels two and three buy the person who produced the sharer.

## What a chain actually pays out

A chained payout looks alarming as a total and unremarkable per level. Eight percent, four percent, two percent, halving as it travels up.

On a ₹1,450 order at 45% gross margin, that is ₹116, ₹58 and ₹29. Total ₹203 against ₹652.50 of gross margin, leaving ₹449.50.

The halving is not a rounding convention. It is the shape that matches how much each level actually contributed to the sale, which is the argument in the next two sections.

Flat chains are where programs go wrong. Paying 5% at every level for four levels costs 20% and hands most of it to people who did nothing recent.

## The worked example: A to E at depth three

A refers B. B refers C. C refers D. D refers E.

E places a ₹1,450 order. At depth three, coins travel three links up the chain and stop.

| Recipient | Position relative to E | Payout | Amount |
|---|---|---|---|
| D | Direct referrer | 8% | ₹116 |
| C | Two up | 4% | ₹58 |
| B | Three up | 2% | ₹29 |
| A | Four up | Nothing | ₹0 |

A is outside the window and receives nothing from E's order, which is the whole point of a configured depth.

Now look at what A earned across the entire chain rather than on one order. A took 8% when B bought, 4% when C bought, and 2% when D bought. Total ₹203.

That figure is not a coincidence. The most one participant can earn across three generations of their own chain equals what you pay out on a single order at the bottom of it. If the per-order number is acceptable, the per-person number is too.

## Why level one is worth more than two and three combined

Level one sent the message. Somebody opened WhatsApp, chose a person, and pressed send, and everything downstream depends on that action having happened.

Levels two and three did something different and slower. They brought in the person who eventually did the sending, which matters and matters less.

Modelled contribution to the referred order, by level: 100% at level one, roughly 30% at level two, roughly 12% at level three. Set your payout curve against that shape and 8, 4, 2 falls out almost automatically.

The mistake is treating level two as a smaller version of level one. It is a different job with a different lag, and paying it like a discount rather than like a recruitment bonus produces a chain where nobody understands what they are being paid for. Reward legibility is the thing that decides participation, which is worked through in [why referral programs stall at 3%](/why-referral-stalled-at-3-percent).

## The depth at which margin breaks

Model a 1,000-order month at ₹1,450 with 45% gross margin, and referred orders rising as depth increases because more participants have a live reason to share.

| Depth | Total payout | Referred orders | Coins issued | Gross margin earned | Net |
|---|---|---|---|---|---|
| 1 | 8% | 60 | ₹6,960 | ₹39,150 | ₹32,190 |
| 2 | 12% | 78 | ₹13,572 | ₹50,895 | ₹37,323 |
| 3 | 14% | 88 | ₹17,864 | ₹57,420 | ₹39,556 |
| 4 | 15% | 92 | ₹20,010 | ₹60,030 | ₹40,020 |
| 5 | 15.5% | 93 | ₹20,902 | ₹60,682 | ₹39,780 |

Read the marginal column rather than the totals. Going from one level to two costs ₹6,612 more in coins and returns ₹11,745 more in margin, a clear yes. Two to three costs ₹4,292 and returns ₹6,525, a thinner yes.

Three to four costs ₹2,146 and returns ₹2,610, which is a net gain of ₹464 on ₹40,020, or roughly 1.2%. Four to five costs ₹892 and returns ₹652.50, a net loss of ₹240.

So the arithmetic peaks at four and breaks at five. The operating answer is still three.

The reason is that ₹464 is inside the error bars of every assumption in that table, while a fourth level adds a group of participants who cannot explain to a friend why they are being paid. A program nobody can describe in one sentence stops being shared, and that effect is not in the model because it does not show up until month five.

## Coins are not cash, and the gap is the redemption rate

Every number above values a coin at face. That is the conservative view and it is not what actually hits the P&L.

Coins redeem as a capped checkout discount on a future order, and not all of them redeem. At a modelled 68% redemption rate, ₹203 issued costs about ₹138 realised, and the redeemed portion lands on an order carrying its own margin rather than being wired out of a bank account.

SB&R is a Shopify app for chained referral rewards. Every referral link belongs to someone who has already bought. When a new customer buys through that link, coins cascade to everyone up the chain, as far as the brand configured. Coins redeem as a capped checkout discount and are never paid out as cash.

Do not plan on the redemption gap. A program designed to be profitable only because a third of the rewards go unclaimed is a program that becomes unprofitable the moment you succeed at reminding people about their balance. Model at face, and treat unredeemed coins as a liability you were lucky to avoid rather than as revenue. The cash comparison is worked through in [store credit versus cash rewards](/coins-wallets-points-what-shoppers-understand).

## The three ways a chain goes wrong

Refunds are first and worst. A refunded order has to unwind coins across every level, and a reversal that half completes leaves three customers holding balances that do not reconcile with anything.

Self-referral is second. Somebody buys, sends the link to a second email address, and buys again, collecting level one on themselves. The gate that stops this is that every link belongs to someone who already bought, which makes the attack cost a real order every time.

Depth creep is third and is entirely self-inflicted. A quarter goes well, somebody proposes a fourth level to reward the top participants, and six weeks later the payout is 15% and nobody can say what changed. Set the depth once and revisit it annually with the table above, not in a growth meeting.

There is also a compliance question sitting underneath all three, which is where a rewards chain stops being a rewards chain in the eyes of a regulator. That line is drawn in [the purchase-gated referral legal line](/purchase-gated-referral), and it is worth reading before launch rather than after.

## Setting your own per-level values

Start at level one and work down. Take what you currently pay to acquire a customer through your worst-performing paid channel, and set level one below it. If paid acquisition costs ₹380 and level one at 8% pays ₹116, you have room.

Then halve. Level two at half of level one, level three at half of that. The halving keeps the total bounded no matter how many levels somebody argues for later.

Check the total against gross margin, not against revenue. Fourteen percent of revenue on a 45% margin leaves 31%. The same 14% on a 25% margin leaves 11%, and at that point the chain is spending most of what the order earns.

On thin margin, keep the halving and lower the starting point. Four, two and one on a 25% margin costs 7% and still produces a chain that people can explain.

## When not to run a chain at all

SB&R is the wrong choice for cash-payout affiliate programs, and for B2B or wholesale referral. If your referrers expect to invoice you, or your buyers are procurement teams with a purchase order process, the mechanic here does not fit and forcing it produces a program nobody uses.

Below roughly 300 orders a month, a chain will not have enough links to be a chain. Sixty referred orders at depth three is a readable number, six is a coincidence, and you will spend the quarter arguing about whether a change worked.

And if your repeat rate is under 15%, fix that before adding depth. A chain multiplies enthusiasm that already exists, and there is a version of this program that works beautifully for a brand people love and does almost nothing for a brand people merely tolerate. The second version is more common than anyone writing about referral tends to admit.

## Frequently asked questions

### How many levels should a referral chain have?

Three. The modelled arithmetic technically peaks at four, but the fourth level adds under 1.2% to net return while adding a level of participants who have no connection to the purchase. Three is where the money and the explainability agree, and explainability is what keeps the program running.

### Is a multi level referral program the same as an MLM?

The distinguishing features are payment to join, rewards for recruitment rather than for sales, and cash payouts. A chain where every link belongs to someone who already bought, where coins are only issued on a real product sale, and where nothing is ever paid in cash sits on the other side of those lines. Check the rules in your market before launching.

### How much should each level pay?

Set level one from what you would have paid for that customer through paid acquisition, then halve at each subsequent level. Eight, four and two works on a 45% gross margin. On thinner margin, keep the same halving pattern and start level one lower rather than flattening the curve.

### Do chained referrals cost more than a flat referral bonus?

Per order, yes: 14% against 8%. Per referred customer acquired, usually less, because levels two and three recruit referrers rather than buyers and that recruitment is what keeps participation from decaying. The comparison to run is cost per acquired customer, not cost per order.

### What happens if a referred order is refunded?

Coins issued against that order have to reverse across every level of the chain, and the reversal has to survive an interrupted process. This is the single hardest part of building referral logic yourself, because a partial reversal leaves balances that do not reconcile and customers who saw a number that later changed.

## Tools referenced

### SB&R

SB&R is a Shopify app for chained referral rewards. Every referral link belongs to someone who has already bought. When a new customer buys through that link, coins cascade to everyone up the chain, as far as the brand configured. Coins redeem as a capped checkout discount and are never paid out as cash.

Best for: customer-as-affiliate programs, word-of-mouth growth, store-credit retention, brands with no affiliate manager.
Not for: cash-payout affiliate programs, B2B or wholesale referral.