Senior director of digital media Paras Shah's read is that an agent only automates a repetitive task, so if waste sits in the system the agent automates the waste faster rather than away. Two years of cutting SSPs and scoring inventory before the bid turns out to be the prerequisite, not a side project.
Why it matters: The cleanest sequencing argument yet for cleaning up and instrumenting the buy before pointing an agent at it.
OpenAI wired AppsFlyer into ChatGPT ads, so installs, in-app purchases and subscriptions finally run through advertisers' own attribution models instead of OpenAI's self-reported numbers. Over 40 brands are testing it, Grubhub among them.
Why it matters: Third-party measurement is the gate on real budget moving to ChatGPT. The CPA benchmarks that follow are the signal, not the integration.
An NYU Stern study is the spine of IAB's new disclosure framework, which triggers a label only where AI creates material deception risk: generated images and video, synthetic voices, digital twins, in-ad conversational agents. Routine post-production, stylized imagery, generic voiceover and copy are exempt. Phase 1 asks for a named AI Disclosure Lead within 60 days.
Why it matters: Disclosure is now a scoping question on any AI creative work. Decide who owns it, which assets trigger it, and what it costs in CTR.
Similarweb opened an AI Ads dataset covering ChatGPT, Google AI Mode and AI Overviews, the three surfaces with no public ad library. Nearly 30% of ad-eligible AI Mode queries show ads and over 40% of Google searches trigger an AI Overview carrying them, placements Google Ads reporting does not break out.
Why it matters: Competitive intel on AI surfaces is now buyable, which turns where are we showing up in AI answers into a reportable line instead of a guess.
Nine months and 51,200 tracked events put average misattribution at 22.4%, or 11,468 sessions logged as Direct instead of Organic, peaking at 29.3% in May. AI Overview traffic hit 16-17% of organic sessions in February-March and has since fallen to 2-4%. Detection comes off the #:~:text= fragment Google appends, caught with a custom GA4 dimension.
Why it matters: A cheap GA4 fix any team can ship this week, and it reframes the organic is down conversation before it starts.
Courtney Johnson repurposed her social content as structured text articles and is logging verified crawlers from OpenAI, Anthropic, Amazon, Apple and Huawei. Colin Rocker ran a citation audit with Claude and found his mentions arriving through LinkedIn and podcasts rather than his own site. On the buy side, Kosterina is using Claude to shortlist partners.
Why it matters: Influencer sourcing is becoming a retrieval problem, which extends AEO work past brand pages into partner discovery.