Conversion-optimized bidding, average daily budgets with intraday pacing, geo exclusions and a Bulk API landed alongside live AppsFlyer and Adjust install and in-app-event attribution across seven markets, with AppsFlyer also routing web events into the Conversions API with dedup. Nothing about the inventory changed; the accountability around it did.
Why it matters: AI-native inventory is now measurable end to end, which makes it a real test line in a media plan rather than a curiosity budget.
Ads Editor 2.13 brings AI Max to Shopping with automated text generation, URL expansion controls, brand lists and URL exclusions, plus PMax retention goals and per-asset AI disclosure toggles for image and video. Operators managing at scale no longer bounce to the web UI to configure the AI layer.
Why it matters: Structured account hygiene starts paying again once the automation can be edited in bulk instead of clicked through campaign by campaign.
An audit of 71 businesses found 17% have no AI-retrievable presence at all, because AI systems extract discrete facts (ownership, location, services, personnel) rather than reading pages. The fixes are unglamorous: named ownership on the homepage, server-rendered core facts, one canonical domain, linked policy pages.
Why it matters: Most sites still fail the basic supply-side checks that decide whether an AI answer can name you at all, and those fixes are cheap.
Ozone data across roughly 20 billion premium impressions shows publisher ad supply down up to 40% in Q2, 32-37% YoY in the US and 39-41% in the UK, with USA Today, Politico, The Economist, Reuters and Reddit all reconsidering terms. Lower volume at higher unit prices rewards impression selection over impression throughput.
Why it matters: Open-web plans need rebuilding around first-party outcomes rather than reach math.
Fastly joined Experian's Agent Trust ecosystem to verify agent identity, delegated authority, intent and payment credentials before requests reach origin. Imperva puts automated traffic at 53%; Cloudflare measured 57.4% of HTML traffic as bot-generated in the week ending June 5.
Why it matters: Conversion-path and attribution assumptions quietly break once authenticated agents transact alongside humans in the same analytics pipeline.
Google signed the EU Code of Practice on AI content transparency, but its own documentation says neither the automatic setting nor a custom label guarantees compliance. Deployers must label synthetic creative prominently at first exposure, with fines up to 15M EUR or 3% of worldwide turnover.
Why it matters: Any EU-facing AI creative needs a labeling step written into the production workflow this week, not after launch.
WARC and LIONS surveyed 400 marketers across four markets: 67% use demographics as the primary input when briefing generative models while 59% say demographic segmentation no longer works, and only 17% always add audience or community insight. Creative volume is up 88%; only 45% report meaningful quality gains.
Why it matters: The constraint on agentic creative output is the brief, not the model, and the brief is still a strategist's job.