Google did not publish a blog post. SEOs still caught two AI Overview interface shifts in the same week: definition queries showing fully expanded Overviews by default (11 November) and a test placing source icons at the bottom of the answer instead of the top (11 November, Search Engine Roundtable).
Neither change alters which pages Google retrieves. Both change whether searchers ever see your favicon before they stop reading.
What changed in the UI
Source icons migrate downward (test)
Historically, AI Overviews often showed a row of cited-site icons above the generated text, with inline links inside the paragraph. Brodie Clark documented a variant where the count of referenced sites moves below the Overview body; hovering reveals the full citation list (SERP Alert).
| Placement | Searcher behavior risk |
|---|---|
| Icons above answer | Sources visible before reading; slightly higher chance of lateral clicks |
| Icons below answer | Answer consumed first; sources become footnotes |
Search Engine Roundtable's framing: bottom placement may make sources harder to find and reduce clicks to publishers — consistent with Pew/ISB zero-click evidence on AI-heavy SERPs.
Google tests UI often; this may not roll out globally. But the direction matches auto-expanded full answers and generative UI eating SERP space: the narrative layer grows; the link layer shrinks.
Definition Overviews expand without a click
Glenn Gabe reported 11 November 2026 that definitional queries now show AI Overviews fully expanded by default — no "Show more" friction (Twitter/X sighting, summarized in Loop Loop).
That extends the August 2026 dynamic expansion Google confirmed for broader query classes. Definitions are top-of-funnel; brands fight for being named in the gloss, not winning the third blue link.
Together, expanded text + bottom icons = more answer, less obvious provenance.
Why this matters for brand visibility — not just CTR
Publishers obsess over click-through. Brand marketers should obsess over in-answer presence:
Mentions vs citations. A brand can dominate the paragraph while a competitor's favicon sits in the footer icon stack — or vice versa. Track both; ghost citations and citation gaps cut both directions.
Zero-click is not zero-influence. Branded search attribution often captures AI-shaped demand better than referral reports when answers satisfy without clicks.
SERP layout drives shortlists. When the Overview is the page, the named brands in text are the shopping list. Favicon position is secondary — unless you are a publisher living on display ads.
UK/EU exposure is already high. SISTRIX measured 80.77% of UK keywords with an AI Overview in early November 2026 (Carlos González weekly). UI micro-tests at that scale still move aggregate behavior.
Same week: other unannounced AIO variants
Loop Loop catalogued additional sightings 10–13 November:
- AI-generated images returning in some Overviews after Google called an earlier test ended (Search Engine Roundtable)
- Local packs inside AIO stripped of call/direction/website buttons — Joy Hawkins called it the "ugliest" local AIO variant yet
- SERP layouts missing the AI Mode tab (unreproduced)
These are layout experiments, not ranking updates. They still change which interactions produce traffic — local businesses may lose click paths even when named.
How this interacts with Google's link policy
In October 2026 Google tightened how inline links in AI answers highlight official sources on sensitive queries (October inline links piece). Bottom icon stacks are a different surface — aggregate citation chrome vs contextual inline trust.
If Google ships bottom icons broadly:
- Inline official-source badges may carry disproportionate trust clicks
- Footer icon hovers become the main discovery path for curious investigators — a smaller audience
- Brand names in prose carry more of the persuasion burden
Measurement playbook
- Do not use favicon rank as a KPI. Track mention rate and citation rate on definitional and commercial prompts separately.
- Screenshot SERPs by device. Bottom-icon tests may be desktop-only or cohort-limited.
- Watch definitional queries in your category — auto-expand increases the chance competitors appear in the first screen without a click.
- Pair GSC generative reports with prompt sampling during volatile weeks (November core update + Search Console logging correction on 9–11 November).
Content and PR implications
Definitions and glossaries: If Google auto-expands definitional Overviews, owned definition pages must state the category truth in the first two sentences — extractability in the first 30% still applies when Google paraphrases you.
Earned authority: When icons sink to the footer, who Google trusts to write the paragraph matters more than who gets the 16×16 favicon. Editorial mentions and structured third-party reviews still dominate citation mix in major studies — 84% earned.
Publisher monetization: If you rely on AI Overview clicks, bottom placement is an adverse trend. Diversify: newsletter, product-led content, brand search capture.
What this does not mean
Not proof Google hides citations maliciously. It is UX testing on a high-zero-click surface.
Not a replacement for ranking work. Classic results still exist below; some searchers scroll. But the default attention is the Overview body.
Not synchronized across countries. UK high AIO penetration ≠ identical UI tests in every locale.
How Obsurfable fits
UI tests do not show up in your CMS. Obsurfable's corpus captures what the answer said and which brands it named across repeated buyer prompts — independent of where Google put the icons. Use Explorer to see whether your mention rate holds when SERP chrome changes; use the App to tie shifts to competitors on the same questions.
FAQ
Should we panic if our favicons disappear from the top?
No. Monitor mentions and citations on a fixed prompt set. Icon position is a CTR lever, not a retrieval verdict.
Do expanded definition Overviews steal traffic from our glossary?
Often they satisfy the query in-SERP. Compete by being the most citable, accurate source Google paraphrases — and by owning adjacent commercial prompts definitions do not cover.
Is bottom placement live for everyone?
As of mid-November 2026 it was a documented test, not an announced global rollout. Treat sightings as signals, not fait accompli.
How does this relate to Hydrogen?
Same week Google upgraded AI Mode's default model (Hydrogen article). Model and UI changes can stack; attribute carefully.
Bottom line
November 2026 AI Overview tests push full answers forward and source chrome backward. For brands, the imperative is unchanged but sharper: win the text of the answer, not the favicon row. Measure mentions and citations on real buyer questions; assume more zero-click reading on definitions and hard queries alike.