← All articles

AI Overviews Click Loss in France: What the Data Shows for Link Builders

By VonixCo Team · · 14 min read

A French SEO analyst's desk with a laptop showing Google search results and a falling traffic graph, illustrating AI Overviews click loss in France

How much traffic did AI Overviews cost French sites? On the five example sites analysed by SEO consultant Glenn Gabe, the answer ranges from almost nothing to nearly 70%. One site lost nearly half its daily clicks, falling from about 750 to about 400 a day. A health-and-finance site lost nearly 69% of its daily clicks, dropping from about 80 to about 25. Two of the five sites held their clicks and click-through rate relatively steady. The honest summary is that AI Overviews click loss in France is real for some sites and invisible for others — and the data, on its own, cannot prove AI Overviews caused the losses.

What happened to clicks when AI Overviews launched in France?

Google launched AI Overviews and AI Mode in France on July 22, 2026. SEO consultant Glenn Gabe then examined Google Search Console data for five example sites, filtering the data to France and comparing clicks, click-through rate (CTR), impressions and generative-AI impressions before and after the July 22 launch. His findings were reported by Danny Goodwin in Search Engine Land on November 9, 2026. Source: Search Engine Land, “Google AI Overviews France analysis” (Nov 9, 2026), by Danny Goodwin, reporting Glenn Gabe’s analysis.

The headline numbers are striking, but they belong to individual sites, not to France as a whole. The first example site saw daily clicks fall from about 750 to about 400 — a drop of nearly 47% — while its CTR fell from 2% to 1.1% and its impressions stayed stable. Impressions staying stable while clicks fall is the detail that matters most: the site was still appearing in search results about as often as before, but searchers were clicking through far less often.

The launch was not a single clean change. The same day, July 22, 2026, Google also launched AI Mode and Search Live in France, alongside a new search box. Several changes arrived at once, so their effects cannot be separated in the numbers alone — a caveat this article carries throughout.

How bad was the AI Overviews click loss in France?

The worst losses in Gabe’s five examples were large. One YMYL site — a site in the “your money or your life” category, covering topics such as health or finance, where Google applies its strictest quality standards — saw daily clicks fall from about 80 to about 25, a drop of nearly 69%. Another YMYL site saw its CTR fall from 1.2% to 0.4%, a decline of about 67%, with clicks falling at launch and then falling again around August 13.

Those are serious numbers for the sites involved, but they must not be read as averages. The sample is five unnamed example sites. No one should quote 47% or 69% as the typical AI Overviews click loss for French sites, because the data does not support that claim — and two of the five sites in the same analysis showed a very different outcome.

The table below summarises all five example sites as reported, so the full picture is visible in one place.

Example site What happened after July 22, 2026
Site 1 Daily clicks about 750 to about 400 (nearly −47%); CTR 2% to 1.1%; impressions stable
Site 2 (YMYL) Daily clicks about 80 to about 25 (nearly −69%)
Site 3 (YMYL) CTR 1.2% to 0.4% (−67%); clicks fell at launch, then again around August 13
Site 4 Clicks and CTR relatively stable despite a sharp rise in generative-AI impressions
Site 5 Clicks and CTR relatively stable despite a sharp rise in generative-AI impressions; traffic had declined before July 22 but recovered by early September

Two of the five sites kept their clicks and CTR relatively stable even as their generative-AI impressions rose sharply. Generative-AI impressions measure how often a site’s content appears in, or is cited by, an AI-generated answer. A sharp rise in those impressions alongside stable clicks suggests the site’s content was being surfaced by the AI feature without losing the clicks it used to get — at least for these two sites, in this period.

One of the five sites also complicates the timeline. Its traffic had already declined before the July 22 launch, and it recovered by early September. That detail is a useful warning against reading any single date as a cause: traffic moves for many reasons, and a decline that starts before a launch cannot have been caused by that launch.

An analytics dashboard showing daily clicks falling sharply while impressions stay flat after AI Overviews launched

Why did some sites lose clicks while others held steady?

The reported data does not say. Five unnamed example sites cannot explain the mechanism. But the pattern fits what search professionals have seen in other AI Overviews studies, so the explanations below are plausible — labelled as such, not proven.

The first explanation is query type. AI Overviews are most likely to satisfy simple, factual questions directly on the results page: a definition, a date, a quick how-to. If a site’s traffic comes mostly from those informational queries, an AI answer can remove the reason to click. If it comes from queries where the searcher needs depth, comparison, tools or a transaction, the AI summary is less likely to replace the visit. The YMYL sites in Gabe’s analysis sit in exactly the informational territory where short AI answers are most useful.

The second explanation is position. A site ranking first used to capture the largest share of clicks for a query. When an AI Overview sits above the first organic result, that top position is pushed down the page, and its click share is the most exposed. A site ranking fifth or sixth may lose less, simply because it had fewer clicks to lose. Our earlier analysis of how Google’s AI features affect clicks covers the same dynamic from a different dataset.

The third explanation is that the AI answer may send clicks as well as take them. When an AI Overview cites a source with a link, some searchers click it — particularly when the question needs more detail than the summary provides. A site whose content is cited clearly may gain some clicks from the AI feature, offsetting the clicks it loses. The two stable sites in Gabe’s sample may have benefited from this, or they may simply serve query types that AI answers do not replace. The data does not distinguish between these possibilities.

The fourth explanation is seasonality and site-specific change. One site’s traffic declined before the launch and recovered by early September, which shows that ordinary variation was happening in the same period. Any single site’s numbers combine the launch effect with its own publishing schedule, its own technical changes and its own seasonal pattern. With five sites, you cannot average that noise away.

Does this prove AI Overviews caused the losses?

No — and this is the most important section of the article. The data shows a correlation in time: clicks fell for some sites after July 22, 2026. It does not, on its own, prove that AI Overviews caused the fall. Three caveats apply, and any honest reading of this study must carry all three.

First, the sample is tiny and unnamed. Five example sites were chosen by one analyst, and we do not know how they were selected, what their topics are beyond the YMYL label on two of them, or whether they are representative of French sites in general. A small, hand-picked sample can show what is possible; it cannot show what is typical. The 47% and 69% figures describe individual sites, not an average, and they should never be quoted as one.

Second, timing alone does not prove causation. Traffic changes for many reasons at once — seasonal demand, a competitor publishing better content, a technical issue, or a different Google update. The site whose traffic declined before July 22 and recovered by early September is the clearest proof of that: its decline had a cause that was not the launch. Without a control group of similar sites not exposed to AI Overviews, you cannot isolate the launch effect from everything else happening at the same time.

Third, the launch was bundled. July 22, 2026 brought AI Overviews, AI Mode and Search Live to France on the same day, together with a new search box. Even if the timing proves something changed that day, the data cannot tell you which of those changes did it. The second drop around August 13 for one YMYL site adds a further puzzle: something changed again three weeks after launch, and the reported data does not say what.

None of this means the study is useless. Five sites is enough to show that large click losses are possible, that they are not universal, and that the pattern — stable impressions, falling clicks — is consistent with AI answers satisfying queries on the results page. What the study cannot do is put a reliable number on the average effect, or prove the mechanism. Treat it as an early warning, not a verdict.

For context, this is not the first dataset to show AI features eating clicks while impressions hold. Our report on Ziff Davis and the decline in organic referrals examined a publisher that reported falling referral traffic in the same period, and the same question — how much of the fall is AI, and how much is something else — applied there too. The France data adds a useful country-level case study, but the same discipline is needed when reading it.

What should link builders do about AI Overviews?

The practical response is not to panic and not to ignore the change either. AI Overviews change what a click is worth and which pages earn one, but they do not remove the value of being the page an AI system cites. The steps below are the ones that follow directly from the France data and from how AI search actually works.

  1. Segment your Search Console data by country and by query type. A total click number hides the story. Filter by the countries where AI Overviews are live, then separate informational queries from transactional and branded ones. If informational clicks are falling while impressions hold, you are seeing the France pattern on your own site.
  2. Check whether your pages are being cited in AI answers. Search your most important queries and look at the AI Overview: does it cite your page, a competitor’s page, or no one? A page that is cited keeps a path to the click; a page summarised without a citation has lost it. Treat a rising generative-AI impression count with falling clicks as a warning sign, not a victory.
  3. Make your pages worth citing, not just worth ranking. AI systems lift short, self-contained answers: a clear definition in the first paragraph, a numbered list of steps, a comparison table, a direct answer to a specific question. Write pages so that a single paragraph or list can be quoted accurately, and make sure the quoted fact is correct. Our guide to AI visibility and the three gates explains how citation, ranking and recommendation work together.
  4. Prioritise pages where the click still matters. Not every page needs to win the click. A page that answers a simple question in one paragraph will lose its click to an AI summary whatever you do — and that is acceptable if the page’s job was awareness, not conversion. Spend your link building effort on the pages where the visitor’s next step is valuable: product pages, service pages, comparison pages, and in-depth guides that a summary cannot replace.
  5. Earn links from pages that AI systems cite. A backlink from a page that appears in AI Overviews carries two kinds of value: the ranking signal it has always carried, and a place in the citation chain that AI search is building. When you choose guest posting targets, ask whether the target site publishes the kind of authoritative, well-structured content that AI systems quote.
  6. Do not chase the lost clicks with thin content. The worst response to a click loss is to publish more short, generic pages targeting the same informational queries — those are exactly the queries an AI answer handles best. Instead, go deeper: original data, a worked example, a framework from your own experience, a comparison only you can make. Content that an AI system cannot generate from existing sources is content it must cite.
  7. Report clicks and citations together, not clicks alone. If you report to clients, add a line for AI visibility: how many of their key queries show an AI Overview, whether their pages are cited in it, and how generative-AI impressions are trending. A client who sees clicks falling without that context will assume the work failed.
Two marketers in a Paris office discussing search strategy over printed charts and a laptop

When should you hire a service like VonixCo?

Hire help when the work above is more than your team can do consistently: segmenting Search Console data by country and query type, checking citation in AI Overviews across your key queries, improving pages so AI systems can quote them accurately, and earning placements on authoritative sites that AI systems actually cite. One page audit is easy; doing it across a whole site, while continuing outreach, takes a team and a process.

VonixCo is a niche-relevant guest posting and SaaS link building service based in Lahore, Pakistan. Its services are SaaS guest posts, niche guest posts, link building and guest post content writing, with plans at $500, $950 and $1,800 per month, content included and a live link report with every placement. Contact: info@vonixco.com and WhatsApp +92 340 0096621.

If your needs are simple — a few pages checked, a few placements earned on sites in your niche — do that work yourself this week. Bring in a service when the volume, the consistency or the outreach grows beyond what your team can sustain. And keep the study’s own lesson in mind: no honest provider can promise that AI Overviews will cite your pages, or that a particular link will recover a lost click. What a good service can promise is relevant placements, honest reporting, and pages that give AI systems something worth citing.

Frequently asked questions

How much traffic did AI Overviews cost French sites?

On the five example sites analysed by Glenn Gabe, losses ranged from almost nothing to nearly 69%. One site lost nearly 47% of daily clicks, a YMYL site lost nearly 69%, and two sites held steady. These are individual site results, not averages for all French sites.

When did AI Overviews launch in France?

Google launched AI Overviews and AI Mode in France on July 22, 2026. AI Mode and Search Live launched the same day, along with a new search box, so the effects of the different features cannot be separated in the data.

What is the source of the France AI Overviews click data?

The data comes from SEO consultant Glenn Gabe’s analysis of Google Search Console data filtered to France, reported by Danny Goodwin in Search Engine Land on November 9, 2026. The analysis covered five unnamed example sites, tracking clicks, CTR, impressions and generative-AI impressions after the July 22 launch.

Did every site in the study lose clicks?

No. Two of the five sites kept their clicks and CTR relatively stable despite sharp rises in generative-AI impressions. One site’s traffic had already declined before the launch and recovered by early September, which shows that factors other than AI Overviews were also affecting traffic in the same period.

Does the France data prove AI Overviews caused the click losses?

No. The study shows a correlation in time, not causation. The sample is five unnamed sites, traffic changes for many reasons at once, and the launch bundled AI Mode, Search Live and a new search box on the same day. The data shows large losses are possible and that stable impressions with falling clicks is consistent with AI answers satisfying queries — but it cannot prove the mechanism.

What should link builders do about AI Overviews click loss?

Segment Search Console data by country and query type, check whether your pages are cited in AI Overviews, and make pages worth citing with clear answers, lists and tables. Focus link building on pages where the click still matters and on sites that AI systems cite, and report AI visibility alongside clicks so clients see the full picture.

The France numbers are an early warning, not a final verdict. Some sites lost nearly half or two-thirds of their daily clicks, and some lost nothing at all — and the difference between those outcomes is the work that matters now: pages worth citing, links from sites worth citing, and honest measurement of both.

Want niche-relevant guest post placements?

Tell VonixCo about your website and niche, and we will reply with suitable publishing options and pricing.

Keep reading