AdRoll opened a pilot giving SMB, mid-market and B2B advertisers serviced access to OpenAI's ad inventory, sitting on top of the self-serve Ads Manager that has been open to US businesses since May. Click-through is still roughly 0.68%, so this is an incrementality test with a support layer, not a scaled channel.
Why it matters: You can test the ChatGPT ad surface through a partner now, without a direct OpenAI relationship or a dedicated buyer on staff.
Croud's index of 2,000+ US consumers found 69% would let AI buy without their approval and 73% research through an LLM before choosing a brand, but 39% cross-check the AI's suggestion against four or more sources first. Brands get eliminated in the verification step, not the transaction.
Why it matters: The strongest data yet that AI visibility work is bottom-funnel, not a brand-awareness line item.
Its MCP server makes account insights, predictive buying stages, 6QA status and keyword intent callable from Claude, ChatGPT, Writer or Agentforce, shipped alongside new APIs and ad targeting that runs on live buying context instead of static account lists. The exported segment is becoming a legacy artifact.
Why it matters: The ABM signal layer is turning agent-callable, which is the plumbing most agent workflows already assume exists.
Cox Automotive's new tracker found 82% of dealers using AI internally but only 29% adjusting strategy for shoppers arriving via AI search. Internal efficiency adoption is running roughly 3x ahead of demand-side adaptation.
Why it matters: A clean, quantified version of the gap in most categories: teams buy AI for their own workflow long before they adapt to how buyers now arrive.
Laura Schiele of Jordan Digital Marketing catalogued 12 gaps operators hit on the platform: no privacy-safe identity targeting, no B2B firmographics, no self-serve incrementality testing, no native CRM integrations. Skai and Innovid completed their Reddit API integrations this week, so enterprise stacks can now buy inventory nobody can cleanly attribute.
Why it matters: Reddit is a heavy LLM citation source, so the paid case has to be argued on visibility rather than platform-reported ROAS.
Aimee Jurenka's five-layer framework separates crawler access, citation visibility, AI referral traffic, dark-funnel demand and business outcomes, on the argument that AI shapes the decision well before the session that gets attributed. Stop hunting for the single AI-search metric.
Why it matters: A defensible reporting spine for AI-search work, where the honest answer is influence across layers rather than one attributable number.