Updated:
Do the Honey and Phia Scandals Mean Affiliate Marketing Is Broken?
Honey and Phia were both caught claiming affiliate commissions on sales they didn't drive, and both were cut off by the networks that policed them. Here's why that's a governance story, not proof the whole category is broken.
Written by the RevU Content Team
No. Two attribution scandals in one corner of a roughly $20 billion industry don't invalidate performance marketing, and in both cases the affiliate networks caught the behavior and cut the offender loose. Honey lost about 7,000 merchant partners and 6 million Chrome users after Rakuten, impact.com, and Awin dropped it; a leaked internal estimate put Phia's misattributed volume at around 51% of the sales it claimed credit for in June 2026. The real story isn't fraud so much as disclosure: both companies grabbed credit for a sale without telling anyone whose credit it was.
The short answer
Same grab, two mechanisms. Honey (PayPal) allegedly overwrote creators' affiliate cookies at checkout. Phia allegedly opened a background tab to override referral codes. Different plumbing, same instinct: both ended up paid for sales they didn't drive, and neither told users that was happening.
The networks worked. Rakuten Advertising terminated Honey on January 12, 2026; impact.com suspended it on January 16; Awin suspended payments on January 21. impact.com suspended Phia in July 2026. That's the enforcement layer of affiliate marketing doing exactly what it's there to do.
The damage was commercial before it was legal. Honey's Chrome install base fell from 20 million to 14 million, its merchant partnerships from ~35,000 to just above 28,000, and its coupon database from ~90,000 codes to ~50,000. A federal judge let the creators' lawsuit proceed on June 22, 2026; nothing has been proven in court.
The category is enormous and mostly fine. Affiliate marketing is roughly a $20 billion industry. These scandals cluster in one narrow layer, browser extensions fighting for last-click credit, not across the whole model.
Offerwalls buy from the same shelf. Much offerwall inventory is CPA offers sourced through the same networks, including impact.com and CJ. The difference is that an offerwall user is told the publisher gets paid, and told exactly what to do to earn the reward.
What did Honey and Phia actually do?
Both were accused of claiming affiliate commissions on purchases they didn't originate, though the two mechanisms worked differently.
Honey. In December 2024, the YouTuber MegaLag alleged that PayPal's Honey extension replaced a creator's affiliate cookie with its own at the moment of checkout, so Honey collected the commission on a sale the creator had actually driven. Reporting later described a system called, with a straight face, "Selective StandDown," that could detect compliance testers by flagging affiliate-network logins, "test" email addresses, and account age, then behave correctly for them while diverting commissions for everyone else. Building a feature specifically to fool your own compliance auditors is a strange way to argue you did nothing wrong. PayPal said the code affected "less than 0.1% of Honey's traffic" in a January 15, 2026 statement.
Phia. In July 2026, Bloomberg reported that Phia, the shopping app co-founded by Phoebe Gates (yes, that Gates) and Sophia Kianni, opened a background browser tab and overrode existing referral codes during checkout, taking credit from other affiliates and from shoppers who'd found the retailer on their own. The company first called it a bug it had learned about "within the last 24 hours." Follow-up reporting on August 11 said the founders had actually known since December 2025, which is a creative definition of 24 hours. An internal Slack estimate on July 7 put the misattributed share at roughly 51% of the gross merchandise value Phia claimed for June, and for a shopping app whose entire pitch is finding you the best deal, giving yourself credit for half your sales is not a rounding error. When the feature was switched off, daily revenue dropped, which tells you how load-bearing that half actually was.
Neither case has been decided anywhere but the press. A federal judge has allowed the Honey lawsuit to proceed to discovery, which is a long way from a finding of liability, and Phia's numbers so far come from leaked internal estimates rather than an audit.
Worth noting, too, that Honey and Phia aren't the only names that have come up this year: researcher Ben Edelman has separately raised questions about influencer platform ShopMy's guidance to merchants on handling duplicate commission claims. It's a different kind of dispute, but it's one more sign that attribution fights keep surfacing across the industry.
Does this mean affiliate marketing is broken?
No, and it's worth being blunt about why, because the "affiliate marketing is a scam" takes are lazy. Affiliate marketing is roughly a $20 billion industry spanning content publishers, review sites, loyalty programs, coupon players, cashback apps, offerwalls, and creator platforms. Two companies misbehaving in one narrow layer of it, the attribution-interception layer where extensions fight for last-click credit, tells you that layer has a governance problem. It doesn't tell you the whole model is rotten.
The best evidence the model works is that the model caught them. Attribution disputes in affiliate marketing don't get resolved by public shaming; they get resolved by the networks sitting between advertiser and partner, the ones who actually enforce the rules of credit. Rakuten, impact.com, and Awin all cut Honey off within nine days of each other. impact.com's CEO, David A. Yovanno, said the company "determined that Honey is out of compliance with our platform policies" and committed to "a software update that will programmatically prevent the type of attribution manipulation we've uncovered." impact.com suspended Phia on the same grounds months later. That's a market policing itself, not a broken one.
What actually connects the two cases is the health of a single convention that most of affiliate marketing depends on: last-click attribution, and the stand-down rules that protect it. A stand-down policy tells extension and toolbar partners not to claim a click another partner already earned. Both Honey and Phia are, at bottom, a story about a company deciding that rule didn't apply to it. Enforce that one convention properly and most of the "crisis" goes away.
Why does this matter to offerwalls and app publishers?
Offerwalls buy from the same shelf. A large share of offerwall inventory is CPA offers, the kind where an advertiser pays for a completed install, a new account, or a finished trial, and that inventory is sourced through the same affiliate networks that just suspended Honey and Phia, including impact.com and CJ. The money runs through a simple chain: an advertiser pays a network, the network supplies offers to an offerwall, the offerwall hands them to a publisher, and the publisher puts them in front of a user. When one link in that chain gets caught manipulating attribution, it's a supply-integrity question for everyone downstream of it, whether or not their users have ever touched a coupon extension.
It also explains why the offerwall model is built the way it is. Offerwalls run on outcome-based pricing: the advertiser pays for a completed action rather than an impression, which is why offerwall eCPMs run far above impression-based formats. RevU publishes a $500+ eCPM figure. That pricing holds up because the action is verifiable. Either the account got created or it didn't; either the in-game milestone got reached or it didn't. There's no last-click argument to have, because the payout is tied to an event someone can watch happen, not to a contested claim about who was standing closest to the checkout button, which is a genuinely undignified thing for a multibillion-dollar industry to be fighting over.
It's also why RevU integrates without an SDK, through an iframe, a hosted page, or a whitelabel API, so a publisher can test the format without shipping a release. The plumbing isn't the interesting part. What matters is that a model built around verifiable completions has far less room for the exact dispute now dragging Honey and Phia through the press and, in Honey's case, the courts. For definitions of the terms used here, CPA, eCPM, last-click, stand-down, RevU keeps a glossary of terms.
What did the coupon extensions get wrong that offerwalls get right?
Disclosure. Both formats monetize attribution; only one of them shows the user the price list.
A coupon extension told the user it was there to find discount codes. Whether it was also claiming a commission, and taking that commission from the creator whose video had sent the user shopping in the first place, was invisible to everyone but the extension itself. Phia's users thought they were comparing prices. Honey's users thought they were saving money. Neither product mentioned that its real business was inserting itself into the attribution chain at checkout, which is a strange thing to leave off the pitch deck.
An offerwall does the opposite by design. The user opts into a clearly labeled transaction: here's the task, here's the reward, and yes, the app gets paid when you finish it. Nobody's confused about who benefits. That doesn't make offerwalls automatically clean. Plenty of them hide task requirements, bury the real payout, or claim attribution before disclosing the task at all, and we've written about those dark UI patterns at length. The claim here is narrower and more durable: the honest version of an offerwall is structurally more honest than the honest version of a coupon extension, because disclosure is the product, not an afterthought bolted onto it.
Coupon extension | Offerwall | |
|---|---|---|
What the user thinks it does | Finds discount codes, get better deal | Pays to complete a task |
When credit is claimed | At checkout, silently | On disclosed task completion |
Does the user know a commission exists? | Usually not | Stated as the reward |
Value to the merchant | Contested | Incremental user action |
Payout basis | % commission | Flat completion bounty |
Coupon sites and cashback apps are turning into offerwalls?
Coupon sites and cashback apps are increasingly borrowing the offerwall's playbook, not the other way around, because a flat bounty for a disclosed, completed action is the easiest version of this business to defend.
Coupon extensions have spent years bolting rewards onto their core product. Honey built its own points program, once called Honey Gold and now folded into PayPal Rewards; Capital One Shopping pays out shopping credits redeemable for gift cards; Coupert stacks coupon-finding and cashback into one tool. It's all the same move, because "we found you a code" is a weak claim to a commission, while "we gave you money back" is a stronger one. Coupon products started looking like cashback products the moment they needed a better story for why they deserved the credit.
Cashback products, in turn, are starting to look like offerwalls. The percentage-of-basket model is giving way to flat bounties for completed actions, especially in fintech, where the reward for opening a new bank account is a fixed payout rather than a cut of a purchase. Open an account, get a set amount. That's not really a coupon, and it isn't really cashback either. It's an offerwall completion bounty wearing a cashback label, whether or not the app wants to admit it, because the economics of "sign up for this" simply don't fit a percentage of anything.
Put plainly, the industry is moving from a percentage of a sale to a flat amount per completed action, because attribution to a purchase is contestable and attribution to a verified action isn't. That's the offerwall thesis, and the rest of the affiliate world is backing into it from the other direction.
Is this an Icarus story?
Partly, but the more useful reading is structural, not moral. Both Honey and Phia had capital structures that demanded a growth rate the honest version of the product couldn't produce.
Honey sat inside PayPal, a public company that paid about $4 billion for it in 2020, a genuinely startling amount of money for a coupon extension even before anyone knew about the scandal, and now has to explain the return on that bet every quarter. Phia was founded in 2025, has raised more than $43 million, sits at a roughly $185 million valuation, and counts a cap table that reads like a Met Gala guest list. Then there's the number that reframes the whole thing: an internal estimate that about 51% of the sales Phia claimed in June 2026 were misattributed, and a revenue drop once that half got switched off. That's not a company with a growth problem. That's a company whose reported growth partly was the practice.
When outside pressure sets the growth number a company has to hit, whether that's a public market watching quarterly earnings or a board expecting a startup to grow into its valuation, the cheapest place to move that number is the measurement layer, not the product itself. It's much easier to reassign credit for a sale that already happened than to generate a new one, which is exactly why attribution is where the pressure lands, not because anyone sets out to commit fraud, but because the spreadsheet doesn't care how the number got there.
RevU's structure is genuinely different here, and it's not incidental color: RevU is 100% privately owned. It answers to no private equity firm, no public shareholders, no venture timeline, and the growth rate it targets is the one the business actually produces. That doesn't make RevU incorruptible; private ownership removes one specific pressure, not the temptation itself, and any offerwall, RevU included, that hides task requirements or grabs attribution before disclosing the task is running a smaller version of the same play. But the pressure that turned a coupon extension's measurement layer into a profit center is a pressure RevU doesn't carry.
