Promo code tracking
Track influencer sales without promo codes or links
A code per creator gives clean counts for the people who type it. Everyone else the post persuaded arrives looking like ordinary sales. Keep the codes, and add a measure that sees the rest.
The short answer
Keep promo codes as a floor: every redemption is a customer that creator certainly brought. Then compare the paying customers who arrived after each creator's posts with your normal level, and judge each creator on that total, with its range, not on redemptions alone. Creators whose codes look weak but whose posts lift sales are exactly the ones a code-only report would wrongly pause.
-
Why codes undercount
To show up in a code report, a viewer has to remember the code, reach a checkout that asks for one, and bother to type it. Many people watch a post, search for your name days later and buy at full price, or subscribe on an App Store screen that has no code field at all. Those buyers are real, and a code report credits them to nobody.
-
What codes are still good for
Codes are the one count you can trust buyer by buyer: a redemption is a customer that creator certainly brought. Use them as a lower bound for each creator, to spot audiences that are ready to buy, and to check that a measured result is at least as high as the customers its codes already confirm.
-
How big the gap is depends on you
The share of buyers who type a code depends on your product, your checkout, the discount and how hard the creator pushes it, so we won't quote a universal ratio. The worked example below shows the shape of the problem: codes confirm a handful of customers per creator, the total is several times higher, and the ranking changes. Your own report shows your own gap.
-
Decide on the total, confirm with the code
Rank creators by the paying customers their posts brought, with the range, and set each against what you paid. Use codes to confirm and to break ties. A creator with few redemptions but a clear rise in sales is a rebook; a creator with no codes and no clear effect is a pause or a cheaper test.
Worked example
Three creators: codes against the measured total
Illustrative numbers, not a real campaign. Each creator was paid US$ 1.500 for the same number of posts.
| Creator | Customers with a code | Paying customers, likely range | Cost per paying customer | Call |
|---|---|---|---|---|
| Creator A | 12 | entre 31 e 49 | entre US$ 31 e US$ 48 | Recontratar |
| Creator B | 3 | entre 14 e 30 | entre US$ 50 e US$ 107 | Testar de novo |
| Creator C | 9 | entre 9 e 13 | entre US$ 115 e US$ 167 | Pausar |
Calls against a cost limit of US$ 60 per paying customer: rebook when the whole cost range is under it, pause when the whole range is over it, test again when it straddles the limit.
By codes alone, C ranks above B. By the total, B brought more than twice as many customers, and C costs too much per customer to rebook.
Questions
Questions, answered.
Should I stop using promo codes?
No. Codes are cheap, they help sales, and every redemption is a confirmed customer. Keep them, and stop treating them as the whole count.
Can't I multiply code redemptions by a fixed factor?
A fixed factor assumes every creator's audience types codes at the same rate. They don't: one audience redeems eagerly, another buys in an app store where there's no code box. Multiplying keeps the ranking the codes gave you, and that ranking is the part most likely to be wrong.
What if a code leaks to coupon sites?
Then the code over-credits that creator, because people who never saw the post use it. Measuring the rise in paying customers after each post, rather than the code, is how you notice: a code count far above the measured range is a sign the code has leaked.
Do I need tracking links or an SDK instead?
No. WhichPost needs a read-only Stripe or RevenueCat key and links to the posts you paid for. The codes you already use appear next to our models' results, so you see both.
Keep reading
Related guides
See what your own creators brought
Connect your revenue read-only and add the posts you paid for, or look through a sample report first. Every result comes with its range.