The Honey scandal changed coupon extensions. Did it fix them?

The awkward part of the Honey scandal was that some of the people telling you to install it were also the people who stood to lose money when you used it. A YouTuber could help you choose a product, send you to the shop and then, according to the allegations, lose the commission when Honey appeared at checkout. Even an unsuccessful hunt for a discount could become a successful hunt for somebody else's referral fee.

That is a much more interesting problem than a coupon code failing. It asks who deserves to get paid for a sale, what the shopper agreed to, and whether a brand's apparently successful marketing is bringing in customers it would otherwise have missed.

And yes, you are reading this on Vouchernaut. A coupon site. The irony comes with the territory. We benefit from people looking for deals, which is precisely why “it recorded a sale” should never be the end of the conversation about whether a deal partner is useful.

Disclosure: Benson has not paid for this recommendation, and these links do not earn us commission.

Honey remains the centre of this story. But by September 2026, the response had reached browser rules, affiliate-network enforcement and a court ruling allowing the creators' case to proceed. The industry has changed parts of the machinery. Whether any particular extension earns its place in your shopping trip, or your shop, still needs a better answer than its own sales dashboard.

The coupon search that became a commission dispute

Honey is a browser extension owned by PayPal. Its familiar attraction is simple: let it try discount codes while you shop instead of searching for them yourself.

On 21 December 2024, investigator MegaLag published Exposing the Honey Influencer Scam. His central allegation was that Honey could replace an existing affiliate referral with its own when a shopper interacted with the extension, including situations where it supplied no saving. He also challenged whether shoppers were being shown the best available codes.

The video examines a purchase journey from a creator's recommendation through to checkout. It is worth watching for that sequence: the shopping basket can look entirely ordinary while the identity credited with the sale changes.

An affiliate link identifies the publisher or creator that referred a shopper. Under a last-click arrangement, the last eligible affiliate interaction can receive the commission. The word eligible matters. A programme can attach conditions to that interaction; being the last thing to happen in a browser does not settle whether those conditions were met.

Consider an illustrative £100 purchase with a £5 referral payment. A reviewer helps someone decide what to buy. The customer clicks the reviewer's link and reaches the retailer. A shopping assistant appears near payment. If the assistant replaces the earlier referral and receives the £5, the retailer can still report a £100 affiliate sale. The customer still receives the product. The reviewer sees no commission.

Nothing in that dashboard alone tells the retailer whether the assistant caused the purchase. It records who received credit under the tracking system. That distinction is the commercial heart of the scandal.

There were two promises to examine

For creators, the concern was losing payment for a customer they had referred. For shoppers, it was whether the helper really was finding the best deal.

ABC's January 2025 reporting covered MegaLag's allegation that Honey could show retailer-approved codes while better discounts existed elsewhere. It also recorded the tension between broad savings claims and terms acknowledging that Honey might not find the best deal. PayPal's response was that Honey followed industry practices, including last-click attribution, and it disputed the allegations.

Those are separate tests. A tool might find a useful discount while raising questions about attribution. It might follow an attribution rule while failing to find the lowest price. “The code worked” cannot answer both.

Nor does finding a code somewhere on the internet establish that everybody is entitled to use it. An offer might be intended for a particular customer, membership or campaign. The meaningful comparison is between offers you qualify for and their final delivered prices. Ten per cent off an inflated starting price can still be a poor deal.

What actually changed after the backlash?

The story did not stop with a viral video. These are the developments that change how to read it now.

Chrome made the benefit part of the rule

Google's policy, updated on 11 March 2025, requires prominent affiliate disclosure and a direct, transparent benefit connected to the extension's purpose. It identifies inserting affiliate links without a discount, cashback or donation as a violation. It also requires a related user action before each affiliate code, link or cookie is included.

That gives shoppers a useful question: what did this interaction actually do for me? Closing a box is a particularly poor moment to discover that a shopping tool regarded you as accepting its involvement in a sale.

There was evidence of a specific Honey change soon afterwards. 9to5Google tested version 17.0.4 and reported that dismissing the “Got it” message in its test no longer replaced the affiliate cookie. That was a meaningful change to the behaviour it tested. It did not settle every later allegation about Honey, every platform, or every purchase journey.

Networks had to address enforcement

In January 2026, Hello Partner reported Rakuten Advertising's notification that it had removed Honey. The network could stop a publisher participating, rather than leaving individual creators to argue over each commission.

Awin's own statement on 21 January said its investigation had confirmed publisher-policy breaches. It had suspended payments and access to new advertiser programmes, with those restrictions still in place at the time of the statement. Awin described a remediation process, including source-code access for verification, handling unauthorised codes and notifying networks about material attribution changes.

An important term here is stand down: restrictions intended to stop a shopping tool intervening when another affiliate already has a qualifying referral. The exact rules vary. A rule on paper also needs a way to detect when it is being ignored.

In its February 2026 explanation of Soft-click, Awin described protection applied by the network itself to reduce the risk of a later extension interaction overriding an earlier publisher. The company says that technology predates this controversy. The useful direction is enforcement built into attribution, alongside expectations placed on extensions.

PayPal responded to the later allegations

The later controversy included allegations that Honey selectively avoided stand-down requirements. In a January 2026 response to Hello Partner, PayPal said it had identified and deactivated the code at issue, that the code predated its acquisition of Honey, and that it appeared to affect less than 0.1% of Honey's traffic. It said it was working with Rakuten on a resolution.

That percentage was PayPal's account of the affected traffic. It was not an independently established measure of creator losses or a court finding. The distinction matters when a small percentage is used to describe activity at a large scale.

The lawsuit survived; it has not delivered a final verdict

On 22 June 2026, Judge Beth Labson Freeman denied PayPal's motion to dismiss the second amended complaint in the consolidated Honey litigation.

At this stage the court was assessing whether the allegations were sufficient for the claims to proceed. The ruling should not be turned into a statement that PayPal has been found liable, that damages have been awarded, or that every allegation has been proved. Equally, an account that leaves the story at an earlier dismissal misses the subsequent ruling.

MegaLag returned to the subject in August with his latest update.

The same questions followed other shopping assistants

Capital One and Microsoft also faced commission-diversion lawsuits; both disputed the allegations, The Washington Post reported in May 2025.

A newer example is Phia. In research published on 11 August 2026, Benjamin Edelman documented affiliate-link behaviour when users dismissed a mobile shopping panel, alongside automatic cookie refreshes. His report includes code and a test recording, and explains why a desktop-only check could miss behaviour on mobile.

Phia's response, as recorded in FMTC's discussion of Bloomberg's reporting, was that it had disabled most revenue-generating features and that an internal estimate overstated the impact because of its methodology. These are separate findings and disputed claims about a different business. They do, however, make it difficult to tell a neat story in which the Honey backlash cured the whole category.

The UK Affiliate and Partner Marketing Association's January 2025 explainer describes how these tools fit into affiliate programmes. Networks and advertisers set conditions for participation, so the agreement governing a particular sale matters as much as the name on the extension.

The sensible standard travels across all of them: deliberate user action, a benefit the shopper can understand, authorised promotions and attribution that reflects the programme's rules.

Yes, a coupon site should be asking this too

A useful discount can make a purchase affordable. A good buying guide can help someone choose the right product or decide to spend less. A cashback offer can give a shopper a reason to choose one retailer over another. Those are things worth paying for.

But a retailer also needs to distinguish those outcomes from paying another intermediary after the customer has already decided. We should welcome that scrutiny even when the intermediary is a voucher publisher. Including us.

There are three useful tests for a deal partner.

• Did the shopper benefit? Was there a usable saving, a useful comparison or an explicitly chosen reward?
• Was credit assigned properly? Did the interaction follow the programme's rules and respect qualifying earlier referrals?
• Did the brand gain something? Did the activity improve sales or contribution after the discount, commission and other costs?

An answer to one is not automatically an answer to the others. A generous discount may delight the customer while costing the shop money on an order that was already coming. An attribution correction may restore a creator's commission without changing the customer's basket at all.

This is where the conversation becomes more useful than “all coupons are good” or “all extensions are bad”.

Why Benson belongs in the conversation

Benson gives retailers a way to investigate extension activity on their own sites. Its proposition is to show detected extensions, observe what happens in those sessions, allow selected pop-ups to be hidden, and compare changes with a holdout: a control group left unaffected by the change being tested.

That is why we recommend starting with its observation tools when a retailer is worried about extensions. The useful outcome could be discovering a problem worth addressing. It could also be finding that extension traffic is too small to deserve much attention, or that a particular arrangement is worth keeping.

Benson also describes attribution reports that flag changes to incoming affiliate or click identifiers, with commission estimates based on the rates the retailer supplies. Those estimates help frame an investigation. An identifier changing is a signal to examine alongside the network's actual transaction records; it does not by itself establish what was paid or what a partner was contractually owed.

The product has distinct jobs. Observe mode shows what Benson would hide. Paid Stop plans add rules for hiding pop-ups. Own adds a retailer-controlled discount widget. The pricing table places code and widget A/B tests on Scale, while blocking holdouts are listed from Stop upwards. Check the BBenson store guideBensonBenson helps online retailers see coupon-extension activity, test pop-up blocking against a control group and offer their own promotions. Start with observation, then choose a paidBuying guide · View store → for the plan distinctions and practical limits.

There is a particularly useful worked explanation in Benson's guide to measuring extension impact. It shows why comparing extension users with everybody else can mislead: those groups may already shop differently. A comparison with a randomly assigned control group can answer a narrower, more useful question about the specific change you made.

Benson's homepage labels its Shopify app as coming soon, and its pricing page applies the same label to checkout code policies. Hiding a visible pop-up should not be mistaken for preventing every background referral change or enforcing which codes a checkout accepts. Check the exact platform, theme and browser coverage before choosing a plan around those features.

The recommendation is to use it to learn what happens in your shop. Buying a blocker and celebrating the number of boxes it hid would simply replace one flattering dashboard with another.

A better test for brands

Start with one question you can act on. For example: does hiding one extension's pop-up improve the money retained per eligible session, after discounts and commissions?

1. Observe before changing the experience. Establish which extensions are detected, where they appear and how often. Review coupon use and actual network payments alongside the sessions.
2. Test one intervention. Keep a randomly assigned control group experiencing the existing setup. If you change the offer, the pop-up and your acquisition campaign together, the result is harder to explain.
3. Count the costs. Revenue per session is useful, but it is not profit. Include discounts once, refunds, commissions, product and fulfilment costs, and the software fee when assessing the commercial result.
4. Give the result room to be uncertain. Small samples and wide confidence intervals can leave a test inconclusive. A positive number on day two is a poor basis for rewriting a partner agreement.
5. Act on what the test actually covered. A result for one browser, extension or offer need not apply to every shopper or partner. Keep, adjust or remove the tested intervention on that evidence.

A blocking holdout estimates the effect of that blocking policy among the sessions it covers. It does not, on its own, measure all the value an extension might provide before a shopper arrives on the retailer's site. That matters if a partner also introduces people to the brand elsewhere.

And if you just want a cheaper basket?

You do not need to become an affiliate-marketing analyst to buy a pair of shoes. Compare the delivered price, check offers you are eligible for, and decide whether the convenience of a shopping extension is worth installing it.

If you want a particular creator to receive a referral payment, a later cashback or coupon interaction may affect that outcome under the relevant programme. Avoid assuming two different services will both receive full credit. Where rewards matter to you, read the retailer's and cashback provider's conditions before combining codes.

A price-history check, a retailer's own promotion or simply leaving the item in the basket for a day may do more for your budget than another extension. If a tool repeatedly produces no useful saving, removing it is a perfectly reasonable response.

The Honey scandal made the invisible part of a familiar transaction visible. The most encouraging response is a higher standard for everyone collecting money from that transaction: show the shopper the benefit, explain the intervention, and let the brand measure what it gained. A coupon site ought to be comfortable with that deal.

Frequently asked questions

Has PayPal lost the Honey lawsuit?

The June 2026 ruling denied PayPal's motion to dismiss the amended creators' complaint, allowing the claims to proceed. That ruling did not establish final liability or award damages.

Did Honey change after the original investigation?

There is evidence of specific changes. In March 2025, 9to5Google reported that the “Got it” interaction it tested in Honey 17.0.4 no longer replaced the affiliate cookie. In January 2026, PayPal said it had deactivated code implicated in later stand-down allegations. These developments concern particular behaviours and dates.

Is Benson an alternative coupon extension for shoppers?

Benson is software for retailers to observe extension activity, test selected interventions and offer their own promotions. Shoppers looking for a discount use the retailer's offers or other shopping services; they do not install Benson as a Honey replacement.

Read NextLEGO Technic for adults: five builds for people who love moving parts