AI and the rise of affiliate fraud is exposing the flaws in the last-click attribution model that affiliates in igaming have been relying on.
Attribution. It seemed to be one of the big talking points at the Affiliate Leaders Summit in Lisbon… and I’m not surprised.
For so long the igaming industry has relied on the last-click attribution model to measure success. It uses third-party cookies to track partner IDs and claim the commission based on what the user does.
But “cookies are not attribution”, Clemence Dujardin, group chief executive of My Affiliates, told delegates – and therein lies one of the challenges. “We need to step back and understand that cookies were never the attribution themselves. We need to better understand attribution.”
The last-click model doesn’t take into account anything the a user hasn’t clicked, but has influenced their decision to convert. And that’s because it was “built for a linear funnel that no longer exists”, Pierre Cadena, chief operating officer at Catena Media tells me afterwards.
“Today a player might watch a creator’s stream, ask an AI agent which casino to try, read a comparison review, and then convert days later on a brand search. Last-click hands all the credit to the final touch and underpays everyone who did the persuading.”
This doesn’t mean affiliate marketing should be abandoned overnight, explains Cadena, as it’s “still the currency that deals settle in”.
“But it should only be one signal, not the whole contract.”
However, Will Barfield, head of fanreach at Better Collective, who sat on the same panel with Dujardin, argues the lack-click model simply “doesn’t work for most”. While Clint Griscti co-founder at apidae.DIGITAL, said the attribution model needs to be “revolutionised”.
That’s partly because AI chatbots like Google’s Gemini or AI Overviews and OpenAI’s ChatGPT are seeing the click-through rates drop. eMarketer’s latest affiliate marketing forecast report showed that as a preferred platform for traffic referrals, Google search fell from 11.1% in June 2025 to just 6.3% in 2026, in North America.
The other challenge facing the industry is all the fragmented igaming and advertising technology and reporting inconsistencies. There is a real lack of standardisation in the industry, Durjadin said, and that includes the definition of what is a conversion.
To move forward, many of the people I spoke to agreed on the need for shared definitions such as of first-party data, cookies and attribution; alongside stronger collaboration among affiliate marketers and even industry-level regulation.
But that poses another question: who should own these standards and hold organisations to account? A question that for now, is probably going to remain unanswered.
For Barfield, a future-proof solution should include something that is “intent-based” and is “more transparent, with full visibility and tracking”. And that’s tracking across the multiple touch points before a user eventually converts.
Add to that, one of the other biggest issues affecting the last-click attribution model is the rise of affiliate fraud, which has cast the net wide open on measuring affiliate performance.
And it is not just the igaming sector. Affiliate marketing has come under scrutiny in the wider industry.
PayPal faced backlash over its Honey browser at the beginning of the year after YouTuber MegaLag accused the publisher of fraud, misleading users or discounts and causing financial loss to creators.
The claims led to three major affiliate networks – Rakuten Advertising, impact.com and Awin – terminating their partnership with the publisher.
Then in July, Phoebe Gates – Bill Gates’ daughter – was accused of affiliate fraud regarding her browser extension, Phia. Gates, alongside her business partner Sophia Kianni, were accused of claiming credit for online sales that they did not actually drive through ‘cookie stuffing’.
So what does the future hold for affiliates?
In the AI era, “measurement has to follow the player, not the click”, says Cadena.
“That means tracking that captures every contributing touchpoint, incrementality testing to prove what actually moved the needle, and deals priced on verified value rather than the last link in the chain.”
It is clear attribution and measurement is lagging behind. But in a world where things are changing faster than ever and an industry that is heavily fragmented, that gap is only going to get wider.
Here’s the good part. Affiliate marketing is still a viable channel and there is data to support that. eMarketer’s most recent forecast suggests that spend on the channel is expected to reach $20bn in the US by 2030.
But for affiliates to survive this wave and not get trapped in the net, they need to solve the attribution problem. Those that focus on that rather than trying to make the last-click model work will be the winners.
