IAB published a measurement standard because more than 20 AI visibility tools return materially different answers for the same brand, and only 16% of brands track it systematically at all. The framework splits "directional" monitoring from "decision-grade" measurement with a nine-criteria bar including a 50-query floor, and pointedly defers attribution.
Why it matters: There is finally a published bar to hold an AI visibility vendor against. If two tools disagree about your brand, the standard tells you which one is even trying to be decision-grade.
Daily caps became seven-day averages the week of July 27, applied automatically to live campaigns, with intraday pacing deciding which days deserve more spend. OpenAI published the weekly total but never a maximum daily overspend multiplier, so nobody can compute worst-case exposure before launch. Click costs have swung as much as 278%.
Why it matters: If you are testing ChatGPT Ads, put a spend guardrail outside the platform this week. The finance conversation now arrives before the performance one.
Attentive, Basis, Salesforce and Alchemer disagree by 4x because they measure different moments: browsing versus buying. Only 35.4% of consumers mostly trust AI recommendations, which is the number that actually explains the gap.
Why it matters: Next time an AI adoption stat lands in your inbox, ask which moment it measured. Discovery numbers and purchase numbers are not interchangeable, and funnels get rebuilt on the confusion.
When buyers hand vendor research to agents, the inputs that decide the shortlist are structured specs, semantic markup, third-party citations and comparison-ready feature matrices. Visual design and traditional SEO stop being the lever.
Why it matters: The work is schema, documentation and citation surface. Those are shippable engineering tasks, not a content calendar.
The "context memory graph" pitch: wire customer data, market signals and the reasoning behind past decisions into one retrievable layer so agent recommendations stay on-brand and auditable instead of starting cold every run. Start with one high-value decision, not the whole funnel.
Why it matters: Most stalled agent projects are starved of retrievable context and decision history, not of model capability. Build the memory layer before you swap models.
CMO Dan Slagen ran a three-phase shift: an outside power user to create urgency, three weeks with coding agents plus five internal coaches, then a build-a-thon where every marketer shipped working software. He personally reviewed roughly 70 video walkthroughs. His line: no tool is coming to save you.
Why it matters: The gap between teams pulling ahead and teams stuck is capability, not tooling. Coaching and a forced shipping deadline move it faster than another platform purchase.