Coupon Affiliate Attribution Calculator

The underlying calculation

The question is how much of a coupon affiliate's attributed volume is incremental: incremental orders = attributed orders × (1 − interception rate).

Interception rate is the share of orders where the customer was already at checkout and left to find a code. It cannot be derived from tracking data and has to be measured experimentally.

The cost of non-incremental orders = intercepted orders × (commission + the discount the code applied).

Worked through

A coupon affiliate credited with 400 monthly orders. A two-week holdout, removing their tracking, shows total orders fall by only 90.

So roughly 78% were intercepted: 310 orders where the business paid commission and a discount on sales it already had.

At $4.82 of commission and a 10% discount averaging $5.80, that is $10.62 per order, $3,292 a month for nothing.

The 90 incremental orders cost the same $10.62 each plus the intercepted total, giving a real acquisition cost of $47 per new customer against a $27 paid CAC.

Where it goes wrong

The holdout has to run long enough to be readable and it costs real revenue if the interception estimate is wrong in the other direction. Two weeks is usually the minimum.

Customers who cannot find a code sometimes abandon rather than buying at full price, and that behaviour is included in the holdout result, which is the correct treatment, since it is a real cost of removing the affiliate.

Making it useful

Run the holdout before deciding anything. Estimates of interception vary from 20% to 80% between businesses and the number cannot be guessed reliably.

Then act on the result: a lower rate for coupon partners, exclusion from last-click credit, or removing the discount code field from the checkout so the search never begins.

The checkout code field as the root cause

A visible discount code box invites the customer to leave and look for one, which is what creates the interception opportunity in the first place.

Removing it, or hiding it behind a link, measurably reduces the behaviour, and businesses that do so frequently find total conversion rises rather than falls, because fewer customers abandon mid-search.

That change costs nothing and addresses the cause rather than paying for the consequence. It is worth testing before any commission restructuring, since it may make the restructuring unnecessary.

Browser extension partners deserve particular scrutiny, since they activate at checkout by design and their entire model is built on the last-click window.

Some networks now allow last-click to be overridden for specific partner types, which is worth asking about rather than assuming the default is fixed.

Running the holdout during a normal trading period rather than a promotional one produces a cleaner result, since a sale distorts both the baseline and the code-seeking behaviour.

Repeating the holdout annually is worth doing, since the partner mix and the browser extension landscape both change faster than most programme terms do.

Restricting code distribution so that only codes issued directly by the merchant work at checkout closes the loop from the other end, since much of the interception depends on codes that were never intended for public circulation.