Key takeaways :
- AI Overview appears automatically above the blue links. AI Mode is a tab the user opens deliberately.
- Both say the same thing 86% of the time, yet cite the same URLs only 13.7% of the time (Ahrefs, December 2025).
- Behaviours invert: in AI Mode, 64% of users click nothing at all; in AI Overview, they re-read and compare before clicking (Clickstream Solutions, 846,000 sessions, 2026).
- Being cited in AI Overview protects the click. Being cited in AI Mode replaces the click. The two require different work.
AI Overview imposes itself on the user, AI Mode is chosen. That interface difference hides a heavier one: the two systems cite different sources 86% of the time.
Does ChatGPT recommend your brand?Measure your presence and identify the brands cited in your place. No credit card required.
What is the difference between AI Mode and AI Overview?
AI Overview is a Gemini-generated summary box, displayed automatically at the top of the results page, above the organic links. The user does not request it: Google decides to show it when it judges the query complex enough.
AI Mode is a separate tab, alongside “All”, “Images” and “Videos”. The user clicks it deliberately. It opens a conversational interface, close to ChatGPT or Perplexity, where questions chain across several turns without returning to a list of links.
That access difference drives everything else. An imposed box slots into a classic search journey, with the blue links right underneath. A chosen conversation replaces that journey.
| Criterion | AI Overview | AI Mode |
|---|---|---|
| Trigger | Automatic, decided by Google | Deliberate, tab clicked |
| Placement | Above the organic links | Separate interface, no blue links |
| Interaction | Read, then back to results | Multi-turn conversation |
| Response length | Short, synthetic | Roughly 4 times longer |
| Top 10 overlap (URL) | 67% | Around 35% |
| Most cited source | YouTube | Wikipedia |
Sources: Semrush (July 2025) for overlap figures, Ahrefs (December 2025) for cited sources.
Read also: How to Appear in Google AI Mode in 2026?
Do both systems cite the same sites?
No. Ahrefs compared both surfaces’ responses across 540,000 identical query pairs. The result: 13.7% shared URLs. Keeping only the top three citations from each response raises the overlap to 16.3%.
The contrast with the substance of those answers makes the figure striking. The same study measures an average semantic similarity of 86%, with 89% of response pairs scoring above 0.8 on a scale where 1.0 means identical meaning. Both systems agree on what to say, and diverge almost entirely on who says it.
Source profiles differ markedly too:
- Wikipedia appears in 28.9% of AI Mode citations, against 18.1% in AI Overview.
- Quora is cited 3.5 times more often by AI Mode.
- YouTube dominates AI Overview citations, which also surface homepages roughly twice as often.
- Health sites are cited about twice as often by AI Mode.
The operational consequence is direct: a brand cited regularly in AI Overviews may be entirely absent from AI Mode on the same queries. Tracking that measures only one surface produces a false picture.
Why do users behave in opposite ways on each surface?
Because one closes the decision while the other opens it. This is the most counter-intuitive finding in the available behavioural data, and the one that changes your strategy most.
Clickstream Solutions analysed roughly 846,000 US Google search sessions in February and March 2026, using anonymised clickstream data. The study measures cursor and scroll movement, not just clicks.
In AI Mode: the closed loop
An earlier study of 185 high-stakes purchases shows that in 88% of AI Mode tasks, users accept the proposed shortlist as-is. They pick the first item listed in 74% of cases. And 64% click nothing at all.
AI Mode operates as a closed circuit. The user reads the answer, chooses from inside the answer, and moves on. If your brand is not in the generated list, it does not exist in that session.
In AI Overview: extended comparison
The behaviour inverts. When an AI Overview is present, cursor positions spread across the equivalent of 83% of the viewport, against 66% without one. More telling, scrolling changes nature: 47.5% of scrolling goes back up the page, against 27% without an AI Overview.
Scrolling back up means re-reading. The user is not scanning, they are comparing. They return to a result already seen, weigh it again, then decide. Your listing on that page gets two to three impressions instead of one, and the second is when comparison happens.
Brand searches lost their shortcut
The most uncomfortable finding concerns navigational queries, where someone types your brand name directly. Without an AI Overview, those users are the most focused group in the study: only 12% remain active at 21 seconds. They know where they are going.
With an AI Overview displayed, cursor activity rises 40% on those same queries, and 45.8% of users are still active at 21 seconds. Someone who typed your brand name now examines what the AI says about you before clicking through.
In my view, this is the data business owners underestimate most. There is comfort in assuming brand traffic is protected because it does not depend on rankings. These figures say otherwise: brand recall no longer works on its own. If an AI Overview shows a comparison where a competitor appears better positioned, some of your brand visitors will see it before reaching you. On the accounts I run, this is now the first thing I check: what Google says about the brand when someone types its name.
Read also: How to Appear in Google AI Mode in 2026?
Should you optimise differently for AI Mode and AI Overview?
The foundation is identical, the differentiating levers are not. Google states this without ambiguity in its official documentation: optimising for generative search is still SEO, and the same fundamentals apply to both surfaces.
That shared foundation covers indexing, snippet eligibility, technical quality and original content. Without it, neither surface retains you.
Beyond that, the two problems are not of the same nature:
Two distinct problems
- Appearing in AI Overview is a comparison problem on the page. Your box, your title and your description get re-read several times. Brand, proof and clarity of positioning decide the outcome on the second pass.
- Appearing in AI Mode is an upstream selection problem. The model builds its shortlist before the user sees anything. What counts is full topical coverage and presence on the sources AI Mode favours.
What weighs more in AI Mode
Depth. With responses four times longer and conversations continuing across turns, a page covering a topic from every angle feeds several successive exchanges. The query fan-out mechanism splits each question into parallel sub-queries: your page earns its place by answering one of those branches.
Presence on encyclopaedic and community sources also counts more, given the weight of Wikipedia and Quora in AI Mode citations.
What weighs more in AI Overview
Video, since YouTube dominates its citations. And work on title tags and meta descriptions, which users now re-read on their second pass over the page.
One point applies to both: Google states it requires no artificial content chunking, no llms.txt file, and no “AI-specific” schema. All three are nonetheless still sold as levers across much of the GEO advice in circulation.
How do you measure presence on each surface?
Separately, otherwise the measurement means nothing. At 13.7% overlap, a combined score blending both hides precisely the useful information.
The Search Console generative AI report is the only first-party data on Google’s side. It shows how your content performs within the generative features of Search.
LLM tracking tools query the engines with your strategic prompts and record your brand’s presence, surface by surface, with competitive comparison on identical prompts.
Google adds a useful warning: no third-party tool has access to its internal ranking systems. A tracking tool observes produced responses; it does not read the algorithm. Any claim of “internal” metrics is marketing.
Disclosure: I am a co-founder of Cockpyt AI, a French tool tracking brand visibility across generative AI engines (ChatGPT, Perplexity, Gemini, Claude). I mention it with first-hand knowledge of what such a tool actually measures, and of its limits.
Does ChatGPT recommend your brand?Measure your presence and identify the brands cited in your place. No credit card required.
Which surface should you prioritise in France in 2026?
AI Overview, for reasons of exposure. Google activated both features in France on 22 July 2026, but they do not reach the same volume of users.
AI Overview displays automatically, with no action from the searcher. AI Mode requires a deliberate click on a tab most French users are only just discovering. The exposure gap between the two surfaces will stay wide for the first months.
Google stated in May 2026 that AI Overviews were used by more than 2 billion people every month worldwide. Comparison with conversational assistants has limits, since the box displays passively, yet it gives the order of magnitude of the exposure switched on across the French market at once.
In my view, operational priority over the next six months goes to AI Overview, without neglecting AI Mode. The reason is calendar rather than technical: AI Overview already produces a measurable effect on your traffic, AI Mode will produce its own once French usage settles. Putting 100% of the effort into AI Mode today means optimising for an audience that has not arrived yet. Putting 100% into AI Overview means ignoring the surface that will take over. A sensible split looks like two thirds, one third.
Read also: Why is Google AI Overview not showing?
Frequently asked questions
Does AI Mode replace AI Overviews?
No, both coexist. Google designed continuity between them: clicking a follow-up question inside an AI Overview switches the user to AI Mode. Both features run on Gemini and draw from the same index, with different models and retrieval techniques.
Can you disable AI Overview on your own searches?
Partially. Some account settings and search operators reduce their display, but Google offers no official global opt-out. On the publisher side, the nosnippet tag prevents your content being used, at the cost of your eligibility for classic snippets.
Does strong organic ranking guarantee citation in both?
No. Overlap between cited URLs and the organic top 10 reaches roughly 67% for AI Overviews and 35% for AI Mode. Ahrefs also measures a drop in AI Overview citation overlap with the top 10, from 76% to 38% in seven months.
Do both surfaces use query fan-out?
Yes, Google confirms it for both. They do use different models and techniques, which explains why the links shown vary between surfaces for the same question.
How do you separate AI Mode traffic from AI Overview traffic in GA4?
Both arrive with google.com as referrer, making the distinction impossible natively in GA4. The dedicated Search Console report remains the most reliable source for measuring performance in generative features.
Should you create separate content for each surface?
No. Google classifies creating distinct pages for every query or surface variation as scaled content abuse. One substantial page, structured in self-contained sections, feeds both systems.
Primary sources
- Google Search Central, “Optimizing your website for generative AI features on Google Search”, updated 10 July 2026, developers.google.com
- Google Search Central, “AI Features and Your Website”, developers.google.com
- Kevin Indig / Clickstream Solutions & Surfer SEO, “Users behave differently in AI Overviews vs. AI Mode”, Search Engine Land, 27 May 2026 (approx. 846,000 Google sessions, February-March 2026), searchengineland.com
- Semrush, “How Google’s AI Mode Compares to Traditional Search and Other LLMs”, July 2025, semrush.com
- Ahrefs, AI Mode vs AI Overviews comparative study, December 2025 (540,000 query pairs; 13.7% URL overlap; 86% semantic similarity)
- Abondance, “Google lance officiellement les AI Overviews en France !”, 22 July 2026, abondance.com
