Between December 2023 and December 2025, the top organic result's click-through rate on queries that later triggered AI Overviews fell from 7.6% to 1.6%. That's not a rounding error. That's a structural shift in how traffic reaches your site from Google Search.
In late August 2026, Google confirmed that AI Overviews can now "dynamically expand" on certain queries, replacing the collapsed snippet with a full-length AI response by default. No "Show more" button. No visible organic results above the fold. Google VP of Product Robby Stein said the expansion triggers only "when the system thinks more detail is useful" and stops once a user starts scrolling. The practical effect: organic listings get pushed further down, and fewer eyeballs reach them.
The Click Math Has Changed Again
The baseline numbers were already ugly. Queries with an AI Overview present show an 8% click rate versus 15% without one. About 1% of visits click the cited sources inside the overview. And 26% of sessions with an AI Overview end on Google entirely, compared to 16% without. Top-ranking CTR drops on affected queries range from roughly 34% to 58% depending on the dataset.
Dynamic expansion makes those numbers directional, not definitive, for the queries it hits. Google didn't disclose what percentage of queries this affects or whether coverage will grow, but the company is also rolling out "generative UI" inside AI Overviews and AI Mode with custom visual layouts, interactive tables, and simulations. The trajectory is clear.
Reddit CEO Steve Huffman put it bluntly: AI Overviews "have not delivered positive value to the broader web ecosystem."
What B2B SaaS Teams Are Actually Doing
The smart response isn't panic. It's measurement, then adaptation.
Standard rank tracking is now insufficient on its own. Teams are tracking AI Overview presence per query, brand citation share-of-voice within those overviews, competitor citations, and CTR deltas between AI Overview and non-AI Overview queries. If you're not measuring these separately from classic SEO KPIs, you're flying blind on a channel changing underneath you.
On the content side, the playbook is shifting from "rank for the keyword" to "get cited in the answer." Common tactics: restructuring pages with question-led headings and short answer blocks designed for extraction, adding FAQPage and Article JSON-LD schema, building topic clusters around entity signals, and publishing original data that's harder for AI to synthesize from generic sources. One Australian meal delivery brand grew its keywords appearing in AI Overviews from 8 to 360 and saw organic purchase conversion improve from 4.48% to 6.42%. ColdIQ reported brand visibility up 105% in AI Overviews with AI referral traffic increasing 169%. A local HVAC company appeared in 71% of tested local queries and attributed a 34% increase in inbound calls to AI-driven discovery.
The Pipeline Diagnostic You Need This Week
Here's the 5-minute version you can run this week:
- Step 1: Pull your top 50 pipeline-generating keywords. Check which ones trigger AI Overviews. Tools like Ahrefs, Semrush, and specialized AI visibility platforms now flag this.
- Step 2: For each affected keyword, note whether your brand is cited in the AI Overview, a competitor is cited, or neither is.
- Step 3: Compare CTR and session-to-pipeline conversion for AI Overview queries versus non-AI Overview queries in the same intent tier.
- Step 4: Quantify the gap. If 30% of your pipeline keywords now trigger AI Overviews and your CTR on those dropped 40%, that's a specific, modelable revenue risk.
The hypothesis: if we restructure our top 10 affected pages with question-led headings, short answer blocks, and FAQPage schema, then our citation rate in AI Overviews will increase because Google's system favors extractable, well-structured content. Success = citation presence on 3+ of those 10 queries within 60 days. Guardrails = no drop in existing organic CTR on non-AI Overview queries. Stop-loss = if organic traffic to those pages drops more than 15% with no offsetting citation visibility, revert and reassess.
The Opt-Out Exists, But Read the Fine Print
Google introduced a Search Console domain-level opt-out for eligible properties to exclude themselves from AI Overviews, AI Mode, and generative AI features in Discover. Worth knowing. Also worth knowing: it doesn't stop model training. Page-level grounding controls aren't scheduled until March 2027. The opt-out is a visibility toggle, not a data protection mechanism. For most B2B SaaS companies, opting out means disappearing from a surface absorbing more search real estate every quarter. That's a trade-off worth modeling, not a reflex decision.
Two years ago, a 7.6% CTR on the top organic result felt like a problem worth solving with better title tags. At 1.6%, the problem isn't your title tag. The SERP itself has been redesigned around an answer that may never send the click.
