The biggest search stories today were not a core update. They were structural. Cloudflare seized control over who gets to train AI on the open web, fresh research showed AI is redistributing search demand rather than erasing it, and two studies exposed how blended metrics quietly hide the readers and revenue slipping away. Here is what moved.
Who Controls the Crawl and the Citation
Two moves shifted the levers over how AI reaches your content and who gets credited for it.
- Cloudflare Will Block AI Training Crawlers by Default From September 15 Cloudflare split crawler permissions into Search, AI Agent, and AI Training, and will block AI training by default across its network unless a site opts in, a change that reaches Google’s own training crawls. Teams behind Cloudflare should audit which category each AI bot falls under before the default flips, or risk cutting citation traffic by accident.
- Proprietary Data Alone Won’t Win AI Citations Kevin Indig argues in Search Engine Land that owning original data is only the entry fee: language models cite whichever page presents a number most cleanly, so an aggregator’s tidy recap can collect the citation your research earned. The fix is structural, front-load the strongest stat and box the methodology, not commission another study.
The Numbers Are Hiding the Story
Two data studies landed on the same warning: aggregate metrics conceal the shifts that matter most.
- A Million-Keyword Study Says Search Demand Is Moving, Not Dying Fractl and Search Engine Land analyzed more than a million keywords and found search volume redistributing by industry rather than collapsing, with FinTech and HealthTech losing while SaaS and Lifestyle gain. The headline 29 percent decline hides which side of that split your vertical actually sits on.
- Publisher Share Data Is Masking a Steeper Under-35 Collapse Similarweb data across 15 UK publishers shows stable younger-audience share concealing a real-terms under-35 decline of up to 34.2 percent, masked by a shrinking older audience. Any team that reports one blended traffic line could be sitting on the same blind spot.
The New Economics of Search Work
One change resets ad costs from the platform side; the other rethinks how the work itself gets priced.
- Google Will Reset CPAs on Budget-Limited Campaigns August 17 Google Ads will start pushing under-delivering Target CPA and ROAS campaigns toward their stated targets on August 17, and a campaign converting at half its target could see cost per acquisition double. A Bid Target Adjustment Tool opens July 6, giving advertisers a six-week window to reset before the change lands.
- Price AI-Assisted SEO Work by Outcome, Not Hours A Search Engine Land column makes the case that once AI compresses a 20-hour deliverable into 20 minutes, hourly billing collapses and outcome-based pricing is what survives. The real risks, accuracy and accountability, do not scale with time spent, so judge the work by whether someone will stand behind it.
Today’s Quick Hits
- Google Retires the AMP Cache Google quietly ended cache-and-viewer serving. AMP clicks now go straight to publishers’ own host pages, so that traffic finally carries the site’s own URL and referrer data instead of a google.com wrapper.