Loading page
Cet article est disponible uniquement en anglais.
Qwairy and Converteo analyzed 12,120 Google AI Overviews answers across France and the US: which queries trigger them, which sources get cited, and why authority and volatility now define visibility.
Google AI Overviews officially launched in France in July 2026. The US has lived with them since 2024. So the obvious question for any brand operating in France: does AI Overviews behave the same way here, and which sources does it actually rely on? To answer with data instead of assumptions, Qwairy partnered with Converteo, a leading French data, AI and acquisition consultancy. Qwairy collected and analyzed 12,120 AI Overviews answers across France and the United States, on 7 industries. The short version: France is already aligned with the US market. And the rules are clear. Brand sites dominate, domain authority acts as a gate, and visibility is reshuffled with almost every answer.
💡 TL;DR
In France, 57% of the sources cited by AI Overviews are commercial sites. The consideration stage (MOFU) accounts for 55% of triggers and is where AI Overviews names the most competitors (4.7 per answer). Sites with a Domain Rating of 70+ are cited 2.7x more than the rest. And for the same query, cited sources change 93% of the time.
As always with AI, results apply to this panel and this period: answers vary from one session to the next.
France and the US look almost identical.
Query type | France | US |
Generic (non-branded) | 62.5% | 62.3% |
Branded | 37.5% | 37.7% |
Short (up to 2 words) | 30.8% | 33.2% |
Medium (up to 4 words) | 45.3% | 43.9% |
Long (more than 4 words) | 23.9% | 22.9% |
Phrased as a question | 29.9% | 29.1% |
Nearly 2 in 3 triggers happen on generic queries, before the user has picked a brand. That is exactly where brands need to be present.
On both markets, the consideration stage drives more than half of all triggers: 55% in France, 54% in the US.
Industry (France) | TOFU | MOFU | BOFU |
Services | 10% | 68% | 21% |
Telecom / Banking / Insurance | 17% | 68% | 15% |
Retail / FMCG | 33% | 52% | 15% |
Luxury & Beauty | 37% | 50% | 13% |
Travel | 30% | 47% | 23% |
Industrial (B2B) | 34% | 40% | 26% |
Energy | 5% | 30% | 65% |
Total | 27% | 55% | 17% |
Services and Telecom / Banking / Insurance reach 68% MOFU queries. These are complex, high-commitment offers (contracts, subscriptions, loans, insurance) that require long evaluation. AI Overviews acts as an instant comparison tool: prices, coverage, customer reviews. Luxury & Beauty (37% TOFU) and B2B (34%) see AI step in much earlier, at the discovery stage: ingredients, routines, standards, technical concepts, well before any purchase decision. Two notable differences with the US:
Lancez un audit gratuit : en quelques minutes, ChatGPT, Gemini et Perplexity répondent en direct à 8 questions d’achat sur votre marché, et vous voyez qui ils recommandent.
# | France | Share of voice | US | Share of voice |
1 | 3.56% | 3.09% | ||
2 | 2.27% | 2.68% | ||
3 | 1.36% | 1.70% | ||
4 | 1.00% | 1.25% | ||
5 | 0.89% | 0.85% | ||
6 | 0.75% | 0.76% | ||
7 | 0.72% | 0.70% | ||
8 | 0.60% | 0.60% | ||
9 | 0.54% | 0.55% | ||
10 | 0.50% | 0.49% |
YouTube ranks first in both countries, but with less than 4% share of voice. No single player dominates: AI Overviews draws on a very wide range of sites. In France, the top 10 is shaped by comparison sites (Meilleurtaux, LesFurets, Selectra, LeLynx) and consumer association Que Choisir, consistent with the weight of consideration queries in banking, insurance and energy.
This is the key finding: 57.2% of the sources cited in France are commercial sites (brands and e-commerce), versus 49.7% in the US. Their share grows as users move toward purchase:
Commercial sites share of voice | TOFU | MOFU | BOFU |
France | 50.7% | 60.2% | 65.6% |
US | 45.1% | 51.7% | 55.3% |
Even at the discovery stage, supposedly purely informational, brand sites capture half of the visibility. For brands, that is good news: your own website remains AI Overviews' first source, as long as it is machine-readable and answers the right questions.
Crossing site type and funnel stage shows AI Overviews switching sources with intent:
French consumers, even at the point of purchase, put real weight on peer opinions. Community discussions are part of your visibility.
Voyez quelles discussions Reddit, YouTube, LinkedIn et forums les réponses IA citent, selon les modèles et pays pris en charge.
Average per answer | TOFU | MOFU | BOFU |
Competitors cited (France) | 1.85 | 4.73 | 1.91 |
Competitors cited (US) | 2.27 | 5.00 | 2.77 |
Sources cited (France) | 13.25 | 12.82 | 11.67 |
Sources cited (US) | 13.88 | 13.88 | 12.98 |
The number of sources stays stable (12 to 13 per answer), but the number of competitors named spikes at consideration: 4.7 brands per answer in France. This is where the visibility battle is fiercest, and where your brand is compared head-to-head.
Sites with a Domain Rating (DR) of 70+ are cited on average 2.7x more than others (3.53 citations vs 1.31 in France, 3.65 vs 1.33 in the US). The median DR of cited sources is 75 in France and 76 in the US. Below DR 70, a site can be cited, but its volume of appearances stays capped. Above 75, the probability of being cited rises sharply. Building domain authority through digital PR and quality backlinks is not a legacy SEO topic: it is the entry ticket to AI Overviews.
YouTube generates the most citations among social platforms in both countries. But Reddit earns a better average position in the source list: when AI Overviews cites a Reddit thread, it often uses it as a primary reference. A useful signal to prioritize community presence and backlink efforts on discussion platforms.
For the same prompt asked 11 times, the share of cases where cited items change from one answer to the next:
Variation rate | France | US |
Sources cited | 92.7% | 96.7% |
Competitors cited | 85.4% | 94.9% |
A single screenshot proves nothing. Being cited today guarantees nothing tomorrow. Measuring AI Overviews visibility requires repeated runs and thinking in probability of presence, not fixed positions. We covered the same blind spot in Same Question, Different Answer .
Sources cited by prompt length | Short | Medium | Long |
France | 4.61 | 1.13 | 1.57 |
US | 4.88 | 1.17 | 1.68 |
Faced with a short, vague query, AI Overviews multiplies sources to cover every angle. Faced with a long, precise one, it identifies the most relevant source with more confidence. The longest queries appear at discovery (6.8 words on average in France), the shortest at consideration (5.8 words). French users are slightly more descriptive than US users at every stage.
The full Converteo x Qwairy report, "AI Overviews & AI Mode in France", goes further: regulatory context, semantic visibility rules, new performance measurement and an operational roadmap for leadership teams. [À AJOUTER : lien de téléchargement de l'étude complète] For tracking both Google surfaces, see Google AI Overviews and AI Mode Launch in France and AI Overviews Ranking Factors .
Suivez les questions qui comptent pour votre activité, voyez quelles marques chaque réponse mentionne et quelles sources elle cite, puis travaillez sur les manques.
Offre gratuite avec 120 crédits initiaux • Sans carte bancaire
Autres articles
The First Answer Isn't the Decision: How AI Assistants Keep Buyers Talking
AI assistants rarely end on an answer: 72% close by inviting another question. And once buyers answer, the shortlist reshuffles. Qwairy's benchmark shows what the next turn decides.
After the Sale, AI Sends Customers Back to Your Website. Until Something Breaks
After a purchase, 90% of AI answers cite the brand's own website, versus 17% before. But troubleshooting is the weak spot. Qwairy's benchmark maps who AI cites at every step after the sale.