The Daily Byte
← All news
PolicyThursday 27 August 2026

The industry is writing the rules for who gets paid when an agent buys

The IAB is drafting a shared framework, due 12th November 2026, for measuring and crediting AI's role in conversions. The reason is blunt: UTMs and referrer data don't reliably survive a customer journey that runs through an assistant. The framework will classify outcomes into categories like agent-initiated, agent-recommended and signal-triggered. Caroline Giegerich, the IAB's VP of AI, is drafting it with a working group of platforms, publishers, agencies, measurement vendors and brands; asked what had been hardest for the group to agree on, she said "everything." The context is already live. Times has started serving ads to AI agents, and Perplexity blocked them, calling them deceptive.
What it actually means
This looks like an advertising story, but it isn't. It's a measurement story, and measurement decides budget. If a meaningful share of your customers now arrive after an assistant did the comparing for them, your analytics is filing those people under "direct". Essentially this means the channel that is actually growing is the one showing zero attributable revenue. Channels with zero attributable revenue get defunded. You will cut the thing that's working because your dashboard can't see it. Two moves before November, neither of which needs a vendor. Ask new customers in the sign-up flow how they found you. The self-reported attribution is crude, but it is the only signal that survives an assistant. Then count agent traffic in your server logs separately from human traffic, so you at least know the size of the thing you can't measure. The reason to do that now rather than wait: Michael Bishop of OpenAds points out the platforms may end up marking their own homework here, the way Facebook built and controlled its own measurement integrations.
12 Novwhen the IAB's AI attribution framework lands
Source: Digiday · Thursday 27 August 2026