Measurement Is Changing Mid-Flight. Marketing Decisions Need to Remember Which Ruler They Used.

U.S. ad buyers are changing how they measure while still deciding on the old measures. The risk is forgetting which ruler each decision used.

2 October 2026 · 4 min read

What happened

IAB published its 2026 Outlook Study: September Update in September 2026 (press release dated 10 September). Alongside a revised spend forecast, it asked buy-side decision-makers how conversational AI tools and agents entering the purchase journey affect the way they measure media.

The survey was sent by email to U.S. buy-side ad investment decision-makers, primarily at brands and agencies (52% agencies, 44% brands), with 211 respondents, fielded 17–31 July 2026. According to the report:

86% are making changes, or expect to within 6–12 months, to how they measure and analyse media performance.
The top measurement challenge, chosen by 45% (up to three answers allowed), is comparing AI-driven with traditional customer journeys.
The most common changes are measuring brand visibility and citations in AI tools (48%), using branded search or direct traffic as proxies (44%) and adding third-party AI-discovery analysis tools (40%).
Only 26% are placing less weight on website traffic, and 23% have direct AI-platform partnerships.

IAB's own summary: "The industry is rebuilding its playbook mid-flight."

As dated context: on 16 September OpenAI began testing sponsored conversations with select U.S. advertisers (reported by The Register), and HubSpot announced ChatGPT Ads with attribution to closed deals. The channel is already being sold. Neither announcement is evidence of effectiveness.

What the evidence actually says

Observed fact: IAB published these results, with a declared method: U.S. only, n=211, email survey, July 2026 fieldwork.
Reported claim: the percentages are respondents' self-reported current or planned changes. They describe intentions, not measured behaviour. The interpretation that there is "a critical lag between strategy and capability", and that buyers "risk making strategic changes based on incomplete visibility", is IAB's.
Inference (ours): most respondents are adding AI-specific measures next to existing website metrics rather than replacing them. So many teams will run two rulers at once, and some will switch rulers between a decision and its review.
Unknown / limitations: whether the plans happen; how brands and agencies differ; whether any of this holds outside the U.S. The sample is ad buyers, not all marketers, and says nothing about Europe.

Record the ruler with the decision, so the review compares like with like.

Why this matters for marketing

A measurement change also changes what a past decision will be judged against.

Take a budget shift made this quarter and justified by growth in branded search, used as a proxy for AI-driven discovery. Two quarters later the team reviews it with a new citation-share metric that did not exist when the call was made. If nobody recorded which measure justified the decision, it can look wrong against a ruler it never used, or right for reasons nobody tested. Either way, the team learns the wrong thing. The survey's top challenge is this comparison problem, reported by buyers themselves.

The operating implication

Two moves for this quarter, and one non-move.

Write down the ruler with the decision. For every material decision: which measures and proxies it rests on, their source and date, and the assumption that links a proxy to an outcome ("branded search growth reflects AI-driven discovery"), marked as untested until something tests it.
Check the ruler before judging the result. At review, first ask whether the measures changed since the decision. If they did, restate the result in the original terms, or label the comparison as not like-for-like.

The non-move: do not rush to replace established metrics. Most surveyed buyers are not doing that either.

The Olymuse lens

This sits under Learning: a team learns only as well as it compares, and compares only what it remembered.

In Your Marketing Team Should Not Have to Start From Zero Every Time we argued that a working memory keeps decisions together with their evidence and assumptions, with source and date, because evidence ages. This signal adds a second way evidence ages: not by getting old, but by being redefined. Our point of view is that the measures a decision relied on are part of that decision's record, not a property of the dashboard.

This is an operating practice any team can start this quarter with a shared document. We are not claiming that software does it for you.

What we would watch next

The comparison gap: whether the next edition of IAB's Outlook study shows fewer buyers citing AI-vs-traditional journey comparison as their top challenge.
Independent measurement: whether platforms selling AI ad inventory publish results advertisers can check outside the seller's attribution.
Shared definitions: IAB released a Measurement Services Addendum v1.0 on 22 September for public comment until 22 October. We will look again after that date at whether it helps contracts state what is measured, and how.

Frequently asked

Does this mean every marketing team should add AI citation metrics now?

No. This is one U.S. survey of ad buyers reporting their own plans: a direction, not a rule. The practice that travels is smaller: whatever you measure, record it with the decision.

Isn't recording the ruler just more documentation?

A few lines per material decision: measures, proxies, the linking assumption, the date. Less than one review spent arguing over a number the decision never used.

Read next: Your Marketing Team Should Not Have to Start From Zero Every Time · Olymuse is in founder-assisted private alpha. Apply for the Private Alpha →