Google AI Overviews and SEO: impact, AI Mode and strategies

This page is for the B2B marketing or SEO manager whose pages now appear in Google results topped by a generated answer. It explains what Google AI Overviews and AI Mode are, how they affect your clicks and SEO, and which criteria lead Google to use a page as a source.

You will know which queries to defend, which pages to rework and how to read a before / after in Search Console.

What Google AI Overviews and AI Mode are

An AI Overview is an AI-generated summary shown at the top of Google results, with links to the pages that support the answer. AI Mode goes further: it answers in conversation, accepts follow-up questions and also cites its sources. Being cited in these two surfaces is a matter of Generative Engine Optimization, covered in the GEO guide.

From SGE to AI Overviews: a rename

SGE (Search Generative Experience) was the name of Google's experiment, renamed AI Overviews in May 2024 when it was integrated into Search, in all markets. "Google SGE" therefore refers to a product that no longer exists under that name: plan your content and your measurement in terms of AI Overviews and AI Mode.

Available in France since summer 2026

At the time of writing, AI Overviews and AI Mode have been available in France since summer 2026. Your French queries can display an AI Overview, and your visitors can switch to AI Mode. Exposure is broad: more than 50% of Google searches display an AI Overview (Squid Impact, 2025). This is a core workstream, not something to merely monitor.

What Google says officially

Google's documentation for site owners sets a sober framework:

• Standard SEO best practice remains relevant, and no special "AI markup" is required.
• A page must be indexed and eligible to show a snippet to be used.
• Being compliant guarantees neither crawling, nor indexing, nor citation.
• The nosnippet, data-nosnippet, max-snippet and noindex directives limit what can be used, AI features included.
• Structured data helps describe a page, without guaranteeing a citation.
• AI-generated content is not penalised as such: Google judges its usefulness.

How an AI Overview builds its answer

To compose an overview, Google may launch multiple related searches across subtopics (a "query fan-out"), then combine information from websites, the Knowledge Graph and community content. This retrieve-then-select mechanism, RAG, is common to all generative engines.

An AI Overview does not extract a whole page. It takes reusable elements: a definition, a fact, a step, a table row. A page split into independent, verifiable blocks supplies more of them than a single long, uniform narrative.

AI Overviews appear mostly on queries that are easy to summarise: simple, informational searches where a synthesis satisfies the user. On complex topics, the synthesis is not enough to decide and the click keeps its place.

The effect on traffic: fewer easy clicks, value that shifts

When the SERP answers, users click less: the overview occupies the top of the page and satisfies part of the intent without a visit. The CTR of position one is 2.6% when an AI Overview is present (Squid Impact, 2025).

Citation partly compensates: being cited as a source in an AI Overview brings +1.08% average CTR (Semrush, 2025). Value moves from ranking to selection as a source: a lower-ranked page that is cited can gain clicks. More benchmarks are gathered in the GEO statistics.

The effect depends on intent, as this table shows:

In B2B, pure informational content is most exposed. Consideration and decision pages are more resilient, because users need proof and a decision framework; there, a citation acts as validation.

SEO does not disappear, its objective changes

Saying AI Overviews "kill" SEO is an oversimplification. Ranking remains the entry point, since the overview relies on indexed, eligible pages. Reaching the top results and then working on citability pays more than gaining one more position: GEO complements SEO by extending its objective to citation.

How to get cited in an AI Overview

Six criteria weigh on whether a page is selected, each with its action:

These extractability criteria are detailed, with a verification protocol, in the article on AI-optimised content.

Structure each section for reuse

Make each section independently citable: a complete answer sentence first, then verifiable elements, then an action or a forward link. The formats that are easy to extract are:

• Q&A: the question phrased like a real search, an answer in two or three sentences, then detail.
• Procedures: listed steps, validation criteria and common pitfalls.
• Comparisons: measurable criteria, assumptions and use cases, with no superlatives.
• Glossaries: a short definition, context, an example and links to related concepts.

Make every claim verifiable

The more easily a claim can be cross-checked, the more likely it is to be reused. Pair each statistic with its source and date, prefer a neutral tone to a promotional one, and show the author, update date and method; the E-E-A-T framework works as a checklist. On sensitive topics (health, finance, legal), state your assumptions and cite official sources.

Check the technical prerequisites

No crawling and indexation means no citations. Check these four points before touching the content:

• Crawling: robots.txt, CDN or hosting blocks, internal links to strategic pages.
• Indexation: accidental noindex, consistent canonicals, duplicate content.
• Rendering: main content available in HTML, verified with URL Inspection.
• Structured data: consistent with visible text (Article, Organization), useful for description but no guarantee of citation.

AI Mode: what changes with conversational search

AI Mode turns search into a conversation: the user asks a question, reads a sourced answer, then chains follow-up questions without leaving Google. Each follow-up is a new chance to be cited, or to be replaced by a competitor.

Compared with an AI Overview, three differences affect your strategy:

• The question replaces the keyword: queries are longer, phrased in natural language, with context.
• The journey continues inside the answer: follow-up questions carry objections, edge cases and comparisons.
• Sources vary from one follow-up to the next: a page cited in the first answer can disappear from the third.

Work on the conversational long tail: anticipate sub-questions, objections and persona-based variants within a single page, rather than multiplying redundant pages. The prerequisites remain those of AI Overviews: indexation and extractable passages.

Measuring the effect of AI Overviews and AI Mode

Google provides no dedicated report. Search Console counts impressions and clicks from AI Overviews and AI Mode in the Performance report, without separating them from classic results. Measurement relies on a page-by-page before / after; the report itself is explained in the Google Search Console guide.

Build a before / after baseline

Without a baseline, the effect of AI Overviews blends with seasonality or competition. Segment your scope first:

• By intent: information, consideration and decision, to compare different CTR dynamics.
• By risk level: easy-to-summarise or complex queries, to know which pages to harden.
• By page: your most viewed pages, each compared with itself before and after.

The clearest signal is CTR falling at a stable position: the page still ranks well, but the overview captures attention. Confirm it with a SERP check. Read any change as a time-based correlation, never as proof of causation.

The KPIs to track

Manage a short chain from visibility to value:

Set the analysis cadence

A fixed cadence lets you decide fast without drowning in data:

• Weekly: alerts for CTR drops at stable positions and for high-impression pages losing clicks.
• Monthly: trade-offs between refresh and new content, prioritisation of pages to harden.
• Quarterly: cluster review, cannibalisation clean-up, recalibration of formats.

Trade-offs: defend, create or do nothing

Treat this like a risk portfolio. Prioritise pages that combine high impressions, easily summarised intent and falling CTR: they finance your organic acquisition. For each, three decisions are possible:

• Defend: the page carries useful traffic; harden it with what a summary cannot condense (decision matrix, detailed comparison, edge cases, checklist).
• Create: the question has no citable answer on your site; publish a page that answers it at the top of a section, with evidence and sources.
• Do nothing: the query is informational, with no business stake, and the overview answers it; take the citation if it comes, without investing in it.

What to do if a page ranks well but no longer gets clicks

First diagnose CTR falling at a stable position in Search Console, then check the SERP for the affected queries: AI Overview present, type of answer, cited sources. Then give a reason to click beyond the summary, depending on the risk:

Finally, strengthen internal links to a decision page: the goal is to protect conversions, not to recover all the lost CTR.

Avoid cannibalisation between similar pages

Give each page a single role: one core intent and one angle. Consolidate repeated content, and turn secondary pages into satellites (use cases, personas, objections) linked to the reference page. An AI Overview finds it easier to select differentiated passages than near-identical pages.

Build the business case for leadership

Base the business case on risk and reallocation, not on volume promises. Structure it in four steps:

• Observation: impacted segments (queries, pages, intent) and before / after baseline.
• Risk: loss of easy informational clicks, pressure from cited competitors.
• Plan: pages to defend or create, citable formats, consolidation of decision pages.
• Measurement: impressions, CTR, conversions and presence in AI Overviews, with a monthly checkpoint.

Depending on your situation

Depending on your starting point, the next step is in another article:

• You want to understand all generative engines: ChatGPT, Perplexity, Gemini and Copilot choose their sources differently from Google, as the article on the AI search engine shows.
• You need to prove your brand's presence: the single set of indicators, from citation per question to referral traffic, is in the article on GEO KPIs.
• You are setting priorities between two disciplines: what they share and where they differ is compared in GEO vs SEO.
• You are getting started: the step-by-step path, from presence audit to measurement, is described in the GEO referencing method.

To track this before / after without multiplying spreadsheets, the Incremys performance reporting module brings together rankings, citations per question and AI traffic, and compares each page with itself before and after an update, as a time-based correlation, without claiming causation.

FAQ: Google AI Overviews and SEO

Are AI Overviews available in France?

Yes, at the time of writing, AI Overviews and AI Mode have been available in France since summer 2026. French queries can display a generated summary at the top of the results, with links to source pages, and users can switch to conversational search.

What happened to Google SGE?

Google SGE was renamed AI Overviews in May 2024, when it was integrated into Search. SGE (Search Generative Experience) referred to the experimental phase, and the name no longer matches a current product. Recommendations written for SGE remain valid when they concern page structure, evidence and indexation.

How do you appear in an AI Overview?

To appear in an AI Overview, a page must be indexed, eligible to show a snippet and well ranked for the query. It must then offer extractable passages: a direct answer at the top of each section, lists, tables, named entities, sources and an update date. No special markup guarantees citation.

Do AI Overviews reduce traffic?

Yes, AI Overviews reduce clicks on informational queries, because the summary answers without a visit. The effect varies by intent: comparisons, methods and decision pages are more resilient. Being cited as a source partly offsets the loss, and the quality of the remaining clicks can be checked in Google Analytics 4.

Does AI Mode replace classic search?

No, AI Mode does not replace classic search: it adds conversational search, separate from the results page. It answers a question, accepts follow-up questions and cites its sources. It is one more surface on which to be cited, with the same prerequisites.

Does Search Console separate AI Overviews and AI Mode?

No, Search Console counts impressions and clicks from AI Overviews and AI Mode in the Performance report, without separating them from classic results. Measurement relies on a page-by-page before / after comparison, tracking CTR at stable positions and SERP checks on critical queries.

How do you measure the impact of AI Overviews on traffic and conversions?

Measure the impact by comparing each page with itself before and after AI Overviews arrived, at comparable positions in Search Console. Then connect changes in clicks to sessions and conversions in Google Analytics 4. Read any change as a time-based correlation, because seasonality and competition also play a part.

What should you do if a page ranks well but no longer gets clicks because of AI Overviews?

A well-ranked page that loses its clicks should be hardened with what a summary cannot condense. First confirm the diagnosis: CTR falling at a stable position and an AI Overview present on the SERP. Then add a comparison, a decision matrix, edge cases or a checklist, and strengthen links to a decision page.