The impact of AI on SEO: what changes, what stays

Search engines now display answers before links, and some searches end without a visit. For a B2B marketing or SEO manager, the question is not whether SEO survives, but where the value of a ranking is moving.

This page sets out the impact of AI on SEO: what changes, what stays, what each workstream needs to adjust, and how to diagnose it in your own data before you decide.

What AI Changes in Search

The main change lies in the interface more than the algorithm: the results page answers by itself, then offers sources. The following table summarises what you observe and how to respond.

These three figures point to the same decision: treat presence in AI answers as an outcome to manage, alongside rankings. Their sources are gathered in our AI statistics, and the mechanism that leads a generative engine to cite you is described in our Generative Engine Optimization guide.

Answer-Led SERPs

At the time of writing, AI Overviews and AI Mode have been available in France since summer 2026. The user reads a synthesis, then may choose a source cited as a link. Competition shifts from "being ranked" to "being selected as a source". Even a good ranking can lose value if the answer satisfies intent before the click: the page must therefore offer what the synthesis does not (evidence, comparisons, data, method).

More Conversational Search

Search is becoming more like a dialogue: longer questions, constraints, business context, and iterations such as "compare" or "for my industry". Your content must answer scenarios, not just keywords: a page that covers a topic end-to-end, in reusable sections, stands out more than a linear text.

Zero-Click, a Risk to Trade Off

Zero-click is not new, but it scales with generative answers. It mostly affects simple informational queries, and far less those that involve a decision. A steering rule lets you decide page by page:

• Intent satisfied in three lines: expect fewer clicks and aim for the citation.
• Intent that requires a decision, evidence or a calculation: aim for the session and the conversion.
• Mixed intent: keep a short answer at the top of the page, then the detail that earns the visit.

What Stays: The SEO Fundamentals

AI shifts value; it does not remove the basics. A generative engine runs a search before answering: a page that is missing from the index or poorly understood has no chance of being cited. Five fundamentals keep their weight:

• Indexing: a page that is crawlable, indexed and accessible to bots, including AI bots.
• Search intent: the page answers the real question, not a keyword variant.
• Authority: links and mentions on reference sites still count, in Google as in AI answers.
• Useful content: Google judges how useful a page is, not how it was produced.
• Measurement: impressions, clicks and conversions remain the foundation, completed by presence in answers.

SEO therefore remains the entry point: without it, work on citations has nothing to stand on.

Which Content Types Are Most Exposed

Queries That Shift to Synthesis

Queries that most often trigger generated answers lend themselves to standardised explanations: definitions, steps, best practices, simple comparisons. That is where zero-click grows fastest. By contrast, as soon as users need specifics (B2B context, constraints, decision-making), demand for sources and depth increases: that is where you keep traffic and gain citations.

Organise Your Portfolio: Reference, Proof, Capture

Part of top-of-funnel demand will be satisfied without a click. Your editorial mix therefore becomes a conversion architecture, which assigns roles per theme across three complementary formats:

• "Reference" pages: the clearest, most complete page (definition, criteria, steps), built to maximise citability.
• "Proof" pages: data, methodology, dated updates, built to maximise trust.
• "Capture" pages: demos, decision-led comparisons, transactional pages, built to maximise conversion.

This split avoids judging a reference page on its sessions: its contribution also shows in the answers that cite it.

What Becomes Critical: Extractability and Trustworthiness

Structure for Understanding and Reuse

Engines and AIs reuse what they can isolate: a stable definition, numbered steps, comparison criteria, a direct answer at the start of a section. Structure matters at least as much as length. To increase extractability without hurting the reading experience:

• start each section with one self-contained summary sentence;
• use lists and tables whenever you compare or describe steps;
• make prerequisites, limits and scope explicit, to prevent misleading summaries;
• standardise definitions, terminology and acronyms across the site.

Improve Trustworthiness

As generative answers become reference summaries, trustworthiness becomes a visibility lever. Original, user-focused content beats keyword stuffing. Use E-E-A-T as a review grid, with verifiable elements:

• visible, consistent update dates;
• external sources cited, with cautious interpretation;
• author pages and clear editorial accountability;
• entities (brand, offers, products, acronyms) aligned across the site.

Reduce Blind Spots

The more you publish, the higher the risk of stacking pages that compete with each other: SEO signals get diluted and AIs match your entities poorly. Each intent deserves one reference page, which satellite pages link back to, and each page must deliver what its title promises.

What AI Changes for Each SEO Workstream

Each workstream keeps its purpose, but its success criterion widens: being ranked remains necessary, being reused becomes an outcome in its own right. The following table gives the adjustment per workstream.

Each row has its reference page: the criteria for content optimised to be cited by AI, the role of links and mentions in AI answers, and the technical and editorial GEO checklist for bot access, indexing and rendering.

Measuring AI Impact With Search Console and Analytics

Diagnose Click Loss

When clicks fall, ask which mechanism is driving the change. At the time of writing, Search Console counts impressions and clicks from AI Overviews and AI Mode in the Performance report, without separating them. Diagnosis therefore relies on a cross-reading:

• Stable position, stable impressions, falling CTR: suspect an interface effect (AI Overview, featured snippet, richer SERP features).
• Falling position, falling impressions: suspect a ranking issue (content, technical, competition).
• Rising impressions, stable clicks: visibility grows without turning into visits, a sign of a more answer-led SERP.

Segment Before You Conclude

Click redistribution varies by query type and page type. Without segmentation, you blend opposing dynamics and draw the wrong conclusion. Minimum segmentation:

• brand vs non-brand queries;
• country, language and device (mobile, desktop);
• directories (blog, resources, product);
• page types (reference, proof, capture).

Connect the Trend to Conversions

Link Search Console and Google Analytics 4, then analyse each organic landing page: direct and assisted conversions, session quality, sessions coming from generative engines. Compare each page with itself before and after a change: the gap is a correlation over time, not proof of cause.

Prioritise Your Pages: The Impact × Effort Grid

The priority is not to fix everything, but to invest where business uplift is most likely. Classify your pages according to the signal found in your data:

Tie each action to an expected outcome: a ranking gain on a profitable cluster, a page that becomes a source in answers, or a greater contribution to pipeline. An action with no named outcome comes off the roadmap.

Move Faster Without Losing Quality

AI speeds up production, but human expertise remains what sets content apart. In B2B, your governance matters as much as your cadence, because an error reused in a generative answer spreads fast. Four rules frame the acceleration:

• a standard brief per page (objective, intent, persona, expected evidence, sources);
• a mandatory factual review (names, figures, scope, dates);
• a cannibalisation check before publishing, to confirm no reference page already exists;
• an update policy for strategic pages, on a defined schedule.

GEO, an Extension of SEO

Faced with these changes, GEO does not replace SEO: it complements it. SEO targets rankings and clicks; GEO targets being cited in the answers of generative engines. Both share indexing, intent and authority, and differ on the unit of work (the question rather than the keyword) and on measurement. How the roles split and how to decide between them is detailed in our GEO vs SEO comparison.

The shift happens page by page: make the content that already carries traffic citable, then measure before / after.

Depending on Your Situation

Depending on where you stand, your next read differs:

• You want to frame the use of AI in your SEO practice: start with our page on SEO and artificial intelligence.
• Your clicks are falling on informational queries: see how to be featured in AI Overviews and AI Mode.
• You need to prove the effect of your actions: rely on the KPIs of visibility in generative engines.
• You are new to analysing your clicks: first master the Performance report in Google Search Console.

To track the effect of these changes on your own pages, the Incremys performance reporting module brings together rankings, traffic from AI engines and citations per question, read as gained, lost, stable or no baseline. Each page is compared with itself before and after publication.

FAQ: The Impact of AI on SEO

What is the impact of AI on SEO?

AI mainly changes click distribution and the value of a ranking. Results pages answer by themselves with AI Overviews and AI Mode, and some searches end without a visit. Ranking remains necessary, but you now also need to be chosen as a source in generated answers.

Will AI kill SEO?

No, AI will not kill SEO: it makes it more hybrid. Generative engines run a search before answering, so a page that is not indexed or poorly understood cannot be cited. What changes is value capture: fewer clicks on simple informational queries, and more competition to become the source that gets reused.

Do AI Overviews reduce traffic?

AI Overviews can reduce clicks on queries whose answer fits in the synthesis, even when the ranking is stable. The effect varies by intent: it weighs on definitions and simple explanations, far less on decision queries. Measure it on your own pages by comparing CTR and position before and after.

Should you change your SEO strategy?

You should adapt it rather than change it. Keep the fundamentals (indexing, intent, authority, useful content), then split your pages between reference, proof and capture. Make reference pages extractable so they get cited, strengthen the decision pages that earn the click, and add presence in AI answers to your measurement.

Is AI content penalised?

No, Google does not penalise content because it was generated by AI: it judges its usefulness. Content produced with AI without review, sources or expertise has little chance of ranking or being cited. Factual review and the input of first-hand experience remain decisive.

Which metrics in Google Search Console help detect a zero-click effect?

Track impressions and clicks together, then interpret them with position and CTR. A zero-click effect shows as stable or rising impressions, a stable position and a falling CTR. Segment by device, country, and brand vs non-brand queries to avoid false diagnoses.

Which pages should you prioritise when visibility rises but sessions stall?

Prioritise first the pages where impressions rise but CTR declines, a sign of a more answer-led results page. Next come pages ranking 4–15 on business-intent queries, then the proof and capture pages that turn qualified traffic into pipeline.