IAB issues AI measurement guidelines, leaves attribution unresolved
Shirley Marschall opens her September 9th ExchangeWire column comparing attribution debates to working out which mosquito bit you on holiday: last-click attribution, multi-touch attribution and marketing mix modelling each answer the credit question differently, and none of them fully accounts for a newer source of influence: AI chatbots increasingly decide which brands even make the shortlist, and that influence can happen inside a conversation that may never produce a click. The IAB has responded with new guidelines for measuring that shift, built around shared vocabulary and quality criteria for vendors that measure "AI visibility," with a full attribution framework left for later. ChatGPT has reportedly reached a $1bn annualised advertising revenue run rate after just six months of carrying ads, according to the column, adding urgency to that gap.
Campaigns measured only by last-click or platform-reported numbers risk crediting an AI-shaped decision to whatever channel happens to catch the final click. A CIMM report cited in the column finds marketing mix models are increasingly wired directly into automated bidding systems, so a skewed allocation can reinforce itself automatically, generating performance data that appears to confirm the original decision. The IAB's guidelines stop at shared definitions for now, and set the vocabulary vendors will need before any AI attribution standard can be built. A separate study cited in the column found attribution is one of the metric areas advertisers expect to matter most over the next three to five years, and the one where confidence in current measurement runs weakest.