Forty-five percent of ad buyers say comparing AI-driven and traditional customer journeys is their top measurement challenge, according to IAB's September 2026 outlook update. And 44% of those buyers are using branded search and direct traffic as proxies for AI influence they can't actually track. That second number is the one that should worry you, because it means nearly half the industry is making budget allocation decisions on a signal they know is wrong.
On November 12, the IAB released its AI-Era Attribution Blueprint. The framework doesn't replace your existing attribution system. It sits alongside it, adding a classification layer that asks three questions about every conversion: who or what created the purchase intent, who or what made the decision, and who or what completed the transaction. Each conversion gets sorted into one of three pathways: Human-Owned, AI-Assisted, or AI-Executed.
Claims Are Not Credit
The sharpest operational idea in the blueprint is the separation between a "claim" and "credit." A claim is eligibility to be evaluated for attribution. Credit is the actual attribution outcome. Claims split further into two types: a decision claim (a party determined the final purchase choice) and a contributory claim (a party helped shape the outcome without making the final call).
Right now, when a vendor or platform tells you a conversion was "AI-driven," there's no shared standard for what that means. Did AI surface the brand during research? Did it compare options and recommend a winner? Did it autonomously complete the purchase? Those are wildly different levels of influence, and the budget implications of each are wildly different too. The blueprint gives teams a taxonomy to tag and sort those claims before anyone starts reassigning credit in a dashboard.
The Evidence Problem Nobody Solved
Here's where the blueprint gets honest about its own limits. Much of the evidence needed to connect an AI recommendation to a downstream purchase is either not captured or locked inside AI platforms. The IAB acknowledges this directly: without improved disclosure and consistent labeling from AI platforms, many influence claims will remain difficult to validate.
As Caroline Giegerich, VP of AI at IAB, put it: "AI-influenced sales are sometimes reported as search or direct traffic, which can send the next marketing investment to the wrong channel."
That mislabeling problem is real and growing. A Bain survey of 1,500 US online buyers found that 44% mostly start their purchase journey in an LLM or split between AI tools and traditional search. DEPT's October 2026 survey of 2,600 US consumers found that 61% had used an AI assistant to research a product or decide what to buy in the prior three months. Among those AI shoppers, 40% used it to discover unfamiliar products or brands and 31% used it to compare options.
Those are consumer numbers, and B2B buying doesn't map one-to-one. But the measurement problem is the same: AI can shape consideration without leaving a clean referral trail. If your buyer researches three vendors through an AI tool, then types your brand name into Google, your analytics logs a branded search visit. The AI influence is invisible.
What You Can Actually Do With This
A classification framework doesn't prove causality. Labeling a journey "AI-Assisted" doesn't tell you the AI interaction changed the outcome or quantify incremental lift. You still need holdouts and experiments for that. But the framework gives ops and measurement teams a practical starting point. Three things worth doing now:
1. Audit your "direct" and "branded search" buckets. What share of those visits might be AI-influenced? You won't have a precise answer, but you can estimate a range and flag it in reporting. The alternative, treating those channels as purely organic intent, is increasingly indefensible.
2. Adopt the claim-vs-credit distinction internally. When a platform or vendor says a conversion was "AI-driven," ask: is that a decision claim or a contributory claim? What evidence supports it? This is a due-diligence filter, not a new dashboard. It costs nothing and prevents premature budget shifts based on unsubstantiated platform reporting.
3. Classify a sample of recent conversions. Pick 50 closed-won deals. Walk each one through the three questions (who created intent, who decided, who transacted). How many are cleanly Human-Owned? How many had an AI touchpoint you can identify? How many you simply can't tell? That last bucket is the measurement gap the IAB is naming, and knowing its size is the first step toward managing it.
The Uncomfortable Part
David Cohen, IAB's CEO, said the blueprint "is honest about what the industry can't yet substantiate." That honesty is useful, but it also means the framework creates more questions than it resolves in the short term. You'll classify journeys. You'll document claims. And for many of those claims, the evidence to validate them won't exist yet because the AI platforms holding the data haven't built the disclosure mechanisms.
This is a vocabulary standard, not a measurement solution. It won't tell you where to move budget next quarter. What it does is give your team, your vendors, and your CFO a shared way to talk about a problem that, until this week, didn't even have agreed-upon terms. For an industry that's been making channel-mix decisions on proxy signals it knows are wrong, a common language is a prerequisite to anything better.