Nine months of GA4 fragment-tracking across 1,661 cited snippets shows AI Overviews drive 7.53% of organic sessions, yet 22.4% of those events are misattributed to Direct. The data also reveals heavy concentration, one snippet drove 2,276 events, and gives brands a concrete playbook to prioritize content refresh.
Google is expanding generative UI beyond AI Mode into AI Overviews, where it now builds custom visual layouts and interactive tools on the fly. Queries like 'ph scale' can return interactive simulations, with full availability in AI Overviews expected within weeks. Sites built around calculators, converters, and comparison tools face fresh competition for direct answers.
Creator briefs now carry a new deliverable: AI visibility. As shoppers start journeys in AI chatbots, LLMs crawl creators' reviews and content to answer queries, so brands like Zoom are hiring creators to build authority they can't claim themselves. Agencies are already tracking creator citations in LLMs, though measurement lags.
New Promptwatch data shows Reddit's share of ChatGPT Search citations fell from 3.83% to 0.52% in mid-August, an 86.4% drop. The timing only loosely matches ChatGPT's shift to site-scoped fanout queries, and Promptwatch itself can't rule out a data-collection issue, so don't hit the panic button on Reddit yet.
A SaaS case study shows brands get dropped from AI recommendations not because of weak authority, but because they fail to publish the evidence AI needs to qualify them for a customer's decision. Stop optimizing for specs; expose the "eligibility gates" that let AI confidently pick you.
Microsoft Clarity's new AI Scrape-to-Referral Ratio in Bot Analytics compares how often AI crawlers hit your site versus send real visitors back, ranked by operator. Microsoft's own example: about 6,000 scrapes per referral. Expect the referral side to be understated, though, since AI platforms often pass no referrer.
Publishers serving ads to AI agents can't prove they worked: no clicks, no standard. Startups like OpenAds (unique referral codes) and Oasy (cost-per-referral via CDN scraping) are racing to build attribution, but most are still proof-of-concept. Measurement remains opaque with no consensus in sight.
IAB released Version 2 of its AI Transparency and Disclosure Framework, drawing the disclosure line around AI that changes authenticity, identity, or what's real, not every use of AI. Realistic synthetic content, digital twins, and human-passing chatbots require labels; routine post-production, copy, and generic voices don't.
Independent agency Butler/Till and iHeartMedia just moved agentic media buying into audio: a four-week, roughly $10K direct buy for an agricultural client, executed by paired AI agents talking over an MCP server. It cut CPMs 42% and pushed 48% of podcast impressions to premium non-skippable mid-rolls, signaling agentic buying is moving from one-off CTV/display tests toward scale.
Starting September 8, qualifying political organizations can join Gmail's Verified Sender Program and reach personal inboxes without standard spam filtering. The catch: senders must pass Campaign Verify checks and keep complaint rates under 0.3% over 14 days. For marketers, it's a live example of Gmail blending sender identity with recipient feedback to grant inbox privileges.
You’re caught up.
That was all 13 stories in the 20 August issue.