800 Code Redemptions Do Not Mean the Post Drove 800 Sales
You are reading a case study that credits a social campaign with sales, and the proof on offer is a number: the discount code in the post was redeemed some hundreds of times. The figure looks solid because a checkout recorded it. Before you repeat it, work out which part the campaign actually earned.
What the case study is actually claiming
Here is a hypothetical. A campaign post offers a discount code, and the report says the code was redeemed 800 times. The 800 is invented for illustration. It is not a reported result from any agency, and it is not from this library.
That sentence makes two claims. The first is that the code was used 800 times. The second is that the campaign caused those uses. The first is a fact recorded by the checkout. The second is an inference the report adds on top. The checkout can tell you an order carried a code. It cannot tell you why the buyer had it.
None of this says the count is false. The checkout recorded real events. The count simply answers a narrower question, “how many orders used this string of text”, than the headline implies, which is “how many sales did the post drive”.
What a redemption count does measure
A code deserves its due. A redemption establishes a completed order, a known discount cost, and a time stamp you can compare with the campaign window. It is cheap, visible and auditable.
It also beats a platform pixel in one respect: it does not depend on cookies or on a click path surviving, so it can catch purchases a click-based report misses. Someone who saw the post on a phone and bought later on a laptop leaves no click trail, but can still type the code.
The error runs the other way too. Buyers who saw the post and did not use the code are invisible to the count, so a code can err low as well as high. This post concentrates on the high side because that is where headlines live.
Four ways a code reaches people who never saw the post
A code is a string of text, and text gets copied.
- Coupon aggregator pages. Some sites collect and list codes, so a code can be found by anyone searching for a discount on that store. It inflates the count because the buyer was already shopping and the post played no part.
- Browser extensions that test codes. Some can try codes at checkout automatically, so a shopper who never saw the campaign can have it applied for them. The discount was never a choice the post influenced.
- Word of mouth and screenshots. A code forwarded in a group chat or forum thread reaches people with no exposure to the original post, so their purchase is counted without the campaign having persuaded them.
- Reuse by existing customers. People already about to buy, or who buy regularly, take the discount because it is there. Their order was coming anyway, the same trap as in revenue growth is not the same claim as new customer growth.
A worked example with hypothetical numbers
Take the invented 800 and split it. The split is made up too, and a real report would need to supply its own.
| Group (all counts HYPOTHETICAL) | Hypothetical count | Can the campaign plausibly claim it? |
|---|---|---|
| Clicked through from the post, then redeemed | 120 | Yes, though some would have bought anyway |
| Redeemed after the code appeared on a listing page | 200 | No, the buyer found the code elsewhere |
| Redeemed after a forwarded message or screenshot | 100 | Partly at best, and it cannot be checked |
| Redeemed by customers who bought in each of the prior months | 380 | No, these orders were likely anyway |
| Total (hypothetical) | 800 |
How many sales did the post drive? Count only the click-through group and the answer is 120. Add the forwarded group generously and it is 220. Count every redemption, as the headline does, and it is 800. Same invented data, and the answer swings by a factor of more than six depending on which groups you admit. The dates matter as well, since a longer count sweeps in more people the post never reached, a point covered in the attribution window is a choice.
The one question that exposes the gap
Put this to any code-based result: how many of these redemptions came from people who could not have seen the campaign, and what did redemptions of that code look like before the campaign started?
The first half forces a count of redemptions outside the campaign’s reach. The second supplies a baseline, so redemptions that would have happened anyway become visible. If the code was already being redeemed at a steady pace before launch, that pace is a floor under the 800, not the campaign’s contribution. It is a cheap approximation of the counterfactual no case study can show you.
A good answer: the code was unique to the post or the creator, published nowhere else, and redemptions were compared against the period before launch. A weak answer: “the code was only shared on social”, with no evidence about where else it appeared. That describes an intention, not what happened.
What a stronger code setup looks like
- Single-use or limited-use codes issued to individuals, rather than one shared public string. A shared string copies freely, while an individual code is hard to pass around.
- A unique code per placement, so redemptions split by post instead of pooling into one number.
- A comparison against a baseline or holdout. The strongest version is in what a trustworthy marketing result requires.
- Disclosure of the discount depth and whether the code was listed anywhere public. A deep discount is more likely to be hunted for.
Reading the next one
When a result rests on a code, ask four things. What was the count? How many codes were there? Where else could the code be found? What did redemptions look like before the campaign began? If the report answers all four, the count means something. If it answers none, you have a record of orders, not of influence.
Ready to practise? Browse the case study library, and if you have published work of your own, submit your own.
FAQ
Is a promo code a reliable way to attribute sales to a social campaign?
Partly. It reliably records that a code was used and when, but not that the buyer saw the campaign, because codes get copied and listed elsewhere. Reliability rises with unique, limited-use codes and a pre-campaign baseline.
Does a code count overstate or understate campaign results?
Either is possible. Leakage and would-have-bought-anyway customers push it up, while people who saw the post and paid without the code push it down. The direction depends on the setup, so ask for the details.
Sources
- The attribution window is a choice, fetched 2026-10-02
- Revenue growth is not the same claim as new customer growth, fetched 2026-10-02
- The counterfactual no case study can show you, fetched 2026-10-02
- What a trustworthy marketing result requires, fetched 2026-10-02