Issue 03

Who really made the sale? Coupon extensions and the attribution problem

13 October 2026

Flicp Editorial

The Indian retail edit desk

Coupon tools at checkout often take credit for sales others created. What the affiliate data, the Honey case and Chrome's new rules show, and why coupons tied to customers keep attribution honest.

A creator in Pune spends a weekend filming a review of a ₹2,499 kurta from a young apparel brand. The video does well. Thousands of viewers tap the link in the description, and a few hundred go on to buy.

At the checkout page, a browser extension pops up on some of those viewers' screens: "We found 3 coupons for this store!" One code works and saves the shopper ₹150. Another click, and the order is placed.

When the creator's affiliate dashboard updates at the end of the month, many of those sales are missing. The brand's dashboard credits them to the coupon extension instead. The creator persuaded the buyer, the brand paid a commission, and the commission went to whoever appeared in the last few seconds.

That story is an illustration, not a reported case. But the pattern behind it is now the subject of a federal lawsuit in the United States, a rewrite of Google's rules for Chrome extensions, and a growing pile of data from the affiliate industry itself. This issue explains how attribution works, where it breaks, and why tying a coupon to a real, signed-in customer keeps the numbers honest, online and at the store counter.

How a sale gets credited

Affiliate marketing pays partners for results rather than reach. A creator, a deals site or a cashback app sends a shopper to a brand, and if the shopper buys, the partner earns a commission, usually a percentage of the order.

The hard part is deciding who sent the shopper. Most programmes still use some version of last-click attribution: when a shopper clicks an affiliate link, a tracking cookie or parameter is stored in the browser, and whichever partner's tag is present at the moment of purchase gets the credit. If two partners touch the same shopper, the later one usually wins.

Last click is simple to run and simple to audit, which is why it has lasted. It also has an obvious weakness. The partner who does the work of persuading someone, often a creator or a review site, is usually early in the journey. The partner who is present at checkout, often a coupon or cashback tool, is late. Under last click, being late is enough.

That gap matters more as the money grows. In India, the IAMAI–Kantar report on affiliate marketing (2023) estimated that spending on the channel was around US$331 million and expected to grow to more than US$420 million in 2025. It put affiliate marketing at around 10–12% of companies' digital marketing spend, against about 15% in some mature markets.

What the affiliate industry's own data shows

If the late partner were also the one who brought the shopper in, last click would be fair enough. The evidence says that often isn't the case.

PartnerCentric, a US partnership marketing agency, published data in February 2025 covering its client programmes from 1 October to 31 December 2024. It compared the conversions that affiliate networks credited to each type of partner with the shopping sessions those partners actually started.

Coupon, loyalty and cashback partners initiated on average only 13% of the conversions credited to them. In the rest, the shopper's session had started somewhere else.
Influencers went the other way: they were under-attributed by 39%, starting more buying sessions than they were credited with.

That's one agency's data from one quarter, and it measures session starts rather than true incrementality. It still points in a clear direction. The partners who close sales are being paid as if they created them, and the partners who create demand are being paid less than they earn.

The IAMAI–Kantar report found the same unease among Indian brands. It noted that many brands find it difficult to judge how effective their affiliate campaigns are and which partners to choose, and worry about fraud. It also put the average affiliate conversion rate in India at around 2%, against a global average of 4%. Lower conversion means every misattributed sale distorts the picture more.

The Honey case

The best-known test of this problem is the litigation over PayPal Honey, a coupon-finding browser extension that PayPal bought in 2020.

In December 2024, a YouTube investigation by the creator MegaLag alleged that Honey replaced creators' affiliate tracking with its own at checkout, including when it found no working coupon. A class action on behalf of creators was filed in federal court in California on 29 December 2024, and several similar cases followed and were consolidated.

The case has moved back and forth since then:

On 21 November 2025, the court dismissed the consolidated complaint without prejudice, finding that the plaintiffs had not adequately shown an injury traceable to Honey or that they were entitled to the commissions.
The plaintiffs filed a second amended complaint in January 2026, attaching actual affiliate agreements with merchants and their commission terms.
On 22 June 2026, Judge Beth Labson Freeman of the US District Court for the Northern District of California denied PayPal's motion to dismiss that complaint, and the case moved towards discovery.

To be clear: these are allegations. Surviving a motion to dismiss means the claims can proceed, not that they are proven, and PayPal has contested them. What the case has already shown is how much attention the market now pays to who claims a sale.

Third-party tracking data reported by PPC Land suggests merchants and users have reacted. Honey's merchant partnerships reportedly fell from roughly 35,000 in December 2024 to just above 28,000, and its Chrome installs from more than 20 million to 14 million by July 2025. Those figures come from independent tracking, not from PayPal.

Google changes the rules for Chrome extensions

Platforms have also moved. On 11 March 2025, Google updated the Chrome Web Store's affiliate ads policy, with enforcement from 10 June 2025. Under the updated policy, an extension may add, change or replace an affiliate link, code or cookie only if:

the affiliate programme is clearly disclosed on the store listing, in the user interface and before installation;
the user takes an action before any affiliate link, code or cookie is applied; and
the affiliate link is tied to a direct benefit for the user at that moment.

Google gave a coupon example: an extension that finds and applies codes must not insert an affiliate link if no coupon or discount is found. Extensions that break these rules can be removed from the store.

That's a meaningful change, but it covers one browser's extension store. It doesn't fix last-click attribution in general, and it says nothing about what happens when a shopper reads a code off a deals site and types it in themselves.

The code itself is the weak point

Most attribution arguments focus on cookies and links. The bigger issue is the coupon code.

A shared code such as FESTIVE20 is a bearer instrument. Whoever holds it can use it, and the code carries no record of who it was meant for or where the shopper found it. A brand might print it in a creator's video description, and within hours it appears on deals sites and coupon extensions' databases. From then on, every order that uses the code looks the same in the brand's reports.

That creates three problems at once:

Wrong credit. The code's original channel loses credit to whichever site or tool repeated it. The creator who negotiated the code may be paid for none of the orders that used it.
Wasted discount. Shoppers who were already going to buy at full price find the code at checkout and use it anyway. The brand gives away margin to customers it already had.
Bad decisions. The brand reads the reports and concludes that coupon sites are its best channel, then moves budget away from the creators and content that actually created demand.

Issue 01 described how leaked codes fuel promo abuse. The same leak that lets fraudsters reuse a code also corrupts the brand's measurement.

Offline is even harder to measure

Online, a brand at least has cookies, sessions and order IDs to argue about. At a store counter, most of that disappears.

When a shopper shows a voucher at a neighbourhood store, the brand usually learns very little: perhaps that a voucher was used, sometimes only when the store submits a claim weeks later. Which campaign brought the shopper in? Which creator? Was it a new customer or a regular? With paper vouchers and shared codes, nobody can say.

That matters in a country where, as Issue 02 noted, more than nine of every ten rupees of retail spending happens offline. Brands are already moving budget into creators, social commerce and performance marketing. If the conversions those channels drive happen in stores and can't be measured, the channels will look worse than they are, and budgets will follow the measurable, not the effective.

Tie the coupon to the customer

The fix for both the leak and the attribution problem is the same: stop issuing codes that work for anyone, and issue offers that belong to someone.

Here is what that looks like in practice:

Issue at the moment of claim. When a shopper claims an offer, the platform creates a coupon for that signed-in customer, not a code that anyone can copy.
Record the source at issue time. The coupon carries the campaign and channel it was claimed from, such as a creator's link, a brand's app or a store's QR code. Attribution is decided when the offer is claimed, not argued over at checkout.
Redeem once and mark it used. At the counter or checkout, the coupon is verified and marked redeemed, so it can't be reused or reposted.
Report from the redemption record. Brand, store and channel partner all read the same record of what was issued, claimed and redeemed.

This is how Flicp coupons work today: each coupon is issued to a signed-in customer and marked used when it is redeemed. A copied code that belongs to someone else doesn't redeem, and an extension can't insert itself between a shopper and an offer they have already claimed.

Identity-bound coupons don't settle every attribution debate. A shopper might see a creator's video, forget about it, and claim the offer later from the brand's own page. Multi-touch journeys are real. But the question changes from "who touched the browser last?" to "where did this specific customer claim this specific offer?", and that question has an answer.

What brands can do this quarter

Audit your codes. Search your shared codes on deals sites and coupon extensions. If a code meant for one creator appears on ten sites, its sales data no longer tells you anything about that creator.
Compare session starts with credited sales. Ask your affiliate network or agency for originated-session data by partner type. If coupon partners are credited with far more sales than they start, reprice them.
Pay for closing and creating differently. Consider lower commissions for late-funnel partners, or paying them only for new customers. Protect creators' credit with unique, per-shopper offers rather than shared codes.
Run a holdout. Pause a coupon partner in one region or for one week and see whether sales actually fall. Incrementality tests settle arguments that dashboards can't.
Measure the counter too. If your offers are redeemed in stores, use coupons that record where they were claimed and where they were redeemed, so offline sales count towards the channels that drove them.

What creators and partners can do

Ask for per-follower offers rather than one shared code, so your audience's purchases stay attributed to you.
Read your agreements. Know how your partner programme defines a qualifying click and what happens when another partner touches the same order.
Keep evidence. Track your own link clicks and compare them with the sales you are credited for. A large, unexplained gap is worth raising with the brand.

What this means for store owners

Attribution sounds like a brand and creator problem, but it reaches the counter. When brands can measure which campaigns bring shoppers into stores, they keep funding those campaigns, and stores keep receiving customers who walk in ready to buy. When they can't, offline offers are the first budget cut.

Coupons tied to customers also protect stores directly. A store that accepts a verified, single-use coupon has a clean record for its claim, which, as Issue 02 explained, is what turns a slow, disputed payout into a routine one.

Five questions to ask about your attribution

What share of your affiliate-credited sales started in a session that partner began?
How many of your shared codes can be found on deals sites right now?
Do your coupon partners get paid when no discount was actually applied?
Can you tell which campaign brought in a shopper who redeemed in a store?
When did you last test whether a channel's sales would disappear without it?

If you can't answer most of these, your channel mix is probably being decided by whoever happens to be last in line.

Coming in Issue 04: why constant discounting trains shoppers never to pay full price, and how limited drops create demand without cheapening a brand.

Sources

The Pune creator is an illustration, not a reported case. The Honey claims are allegations that have not been decided on their merits.

This newsletter is general information, not legal, financial or tax advice. Product details describe Flicp as it works today and may change.