Fractl's AI visibility index quantifies the growing gap between traditional search authority and AI recall: brands that dominate Google can barely surface in AI answers, while smaller players over-index. Notable outliers: Lemonade outranks State Farm in insurance, and sustainability-coded DTC brands beat retail giants in lifestyle. Traditional SEO authority still matters, but it's no longer the whole map.
Google's Quality Rater Guidelines define effort not as hours worked but as visible evidence of human-created, original value, and the API leak exposed contentEffort and ugcDiscussionEffortScore signals. For brands, that means scaled AI filler is algorithmically punished, and firsthand research is the ranking edge.
Google's performance report quietly records AI Mode chat fragments ("yes go on") as queries, leaking conversation data Google's new Generative AI report omits. A practitioner pulled 1,127 of them from 16 months of data, sorted them into seven buckets, and shipped a free classifier so you can mine your own. This is how you see the floor of AI visibility Google won't expose.
Analysis of 300M+ US searches shows Google's AI Overviews cite Facebook 19.5M times a month, Instagram ~877K, TikTok ~78K. AI rewards content that answers the exact query, not big audiences: a 3,000-follower post outranks a 500K-follower brand. Reach is not influence, so audit which sources AI actually pulls.
The AI-bot blocking and licensing fight is aimed at Google, but DataDome data shows Meta's crawlers now drive the majority of AI agent traffic, returning almost nothing in referrals while Google's bots stayed the most-blocked. The author argues publishers are negotiating over a shrinking traffic promise that Meta never made in the first place, leaving a licensing-only outcome as the likely end game.
The ANA's Retail Media Measurement Standardization report gives brands a checklist for comparing retail media networks: third-party validation, transparency into methods, a fixed 14-day attribution window, and one shared definition of "outcomes." Built with 40+ marketers and feedback from Walmart, Target, CVS and more, but not Amazon, it aims to end "purple to bananas" comparisons.
Georgia-Pacific's digital media lead Paras Shah says the tech is ready but programmatic isn't: drop an agent into a messy system and you automate the waste faster, not away. So the brand is cleaning up supply paths and inventory scoring first, splitting adland into a cautious brand camp versus holdcos pushing agent-to-agent buying.
Helena, Enrich Labs' autonomous marketer, independently picked Times Square, wrote the self-aware copy, cold-emailed media vendors, and negotiated the rate from $25K to $10K a day. It's the clearest example yet of an agent that doesn't just automate, it executes a real media deal end to end.
You’re caught up.
That was all 13 stories in the 19 August issue.