Search engines still dominate the search for commercial information
When looking for information on a product or service before making a purchase, the vast majority of French people continue to rely on long-established reflexes. Search engines remain the central tool: 64% of French people cite them among their two main means of information, 44% of them first. They are followed by word-of-mouth (52%) and in-store advice (40%), which continue to play a key role in the purchasing decision. In contrast, conversational agents with artificial intelligence still appear to be marginally used: only 10% of French people cite them as a means of obtaining information about a product or service, and barely 4% make them their first reflex. In terms of the population as a whole, AI is not yet competing head-on with search engines in the pre-purchase phase. However, this observation masks a revealing paradox. Nearly one French person in two (48%) has already used a conversational agent, and a majority consider them more effective than search engines for understanding a subject (64%), finding factual information (61%) or obtaining personalized advice (58%). In other words, the French use AI and consider it effective… but don’t spontaneously think of it when they want to buy. This discrepancy between favorable perception and lack of reflex is the main obstacle to be overcome if AI is to become a structuring channel for commercial research.
Among everyday users, AI is already nipping at the heels of search engines
Behind this national average lies a more contrasting reality. If we zoom in on daily users of AI tools (i.e. around 7% of the French population), the landscape changes radically. In this segment, AI is cited by 24% of users as the primary pre-purchase information channel, compared with 33% for search engines. The gap, while still real, has narrowed considerably: AI is no longer a marginal channel, but is nipping at the heels of search engines. In cumulative quotes, conversational agents even reach 44% among these intensive users, a level comparable to word-of-mouth (38%) and much higher than sales advice (24%). These 7% of French people are not an anecdotal niche. They represent the early adopters, whose behavior generally prefigures mass use. And for them, the switch to AI as a sales research tool is not a prospect: it’s already a reality.
Rapid diffusion driven by younger generations
This situation is part of a context of accelerated adoption of artificial intelligence tools. Nearly one French person in two (48%) has already used at least one conversational agent, a proportion that rises to 82% among 18-24 year-olds, compared with just 20% among those aged 65 and over. Higher education graduates (64%) and residents of the Paris metropolitan area (59%) are also clearly over-represented. Among users, the frequency of use is sustained: 59% use these tools at least once a week, and 14% on a daily basis. Conversational AI is becoming an everyday tool for a growing proportion of the population. This massive penetration rate among the younger generations augurs even stronger adoption over the next few years. By cohort effect, the proportion of regular users in the total population is set to grow mechanically, as the generations familiar with these tools take on a larger role in the adult population. Among the tools available, ChatGPT stands out as the hegemonic player: 91% of AI users claim to have already used it, far ahead of Gemini (43%), while Perplexity, Claude or Deepseek remain confidential (around 10% or less). However, this concentration lessens among intensive users: among daily users, Claude reaches 23% penetration, while Perplexity and Deepseek are around 18%. Power users are more demanding, querying several AIs and comparing their answers.
Substitution for Google already underway for part of the search market
The substitution movement is already significant on a population scale. 45% of French people say they have replaced at least part of their Google searches with conversational agent searches. Even more notable: 21% of French people (i.e. more than one in five) say they have done so for almost all or a significant part of their searches. This is no longer an exploratory or occasional use: for one-fifth of the population, AI has already profoundly altered online search habits. How, then, can we explain the apparent gap between these 45% substitutes and the mere 10% who cite AI as a source of information before making a purchase? The answer may lie in the nature of the searches substituted. The French first switched to AI for those uses where it is perceived to be most effective: understanding a subject, obtaining an explanation, seeking inspiration. These are informational, educational and creative searches, not transactional ones. AI has conquered the field of general information, but not yet that of commercial information. This reading is consistent with the hierarchy of perceived effectiveness: AI’s comparative advantage erodes precisely the closer we get to the purchase. Among AI users, on the other hand, the rate of substitution is becoming the norm: 88% say they have replaced at least part of their searches, and this figure rises to 95% among daily users. Among the latter, almost half (47%) have switched almost all their searches. Generational dynamics amplify this finding: 73% of those under 35 say they have made at least a partial substitution, compared with only 18% of those aged 65 and over. Replacing Google with AI is becoming the default behavior of younger generations.
Perceived superior effectiveness for information, more nuanced as purchase approaches
Compared to traditional search engines, conversational agents are generally perceived as more effective for informational uses. Nearly two-thirds of French people believe they are more effective for quickly understanding a subject or news item (64%), finding factual information (61%), or getting ideas and inspiration (59%). This perceived superiority is due to their ability to synthesize and contextualize, which meets a need for rapid comprehension. On the other hand, as soon as the search approaches the commercial act, the comparative advantage fades: only 52% consider AI to be more effective for preparing a purchase, and 43% for finding a service provider, the only use where search engines retain a clear perceived advantage. However, this hierarchy changes among regular users. Among weekly users, 74% consider AI more effective for preparing a purchase, and 85% for obtaining personalized advice. Repeated experience gradually lifts the brakes: what appears to be a structural limitation in the general population looks more like an adoption lock that is bound to give way as usage spreads.
Conclusion: two France’s, one trajectory
The survey reveals the coexistence of two information-seeking regimes. In terms of the total population, search engines retain a dominant position. They remain the first reflex for 44% of French people before a purchase, compared with 4% for AI. And while 45% say they have substituted some of their searches, this substitution is primarily for informational uses (understanding, learning, getting inspired) rather than commercial ones. For the majority, the switch to AI as a commercial search channel is still a long way off. Among regular AI users, and particularly among the 7% of daily users, the situation is radically different. Here, AI is right on the heels of search engines as the leading channel for commercial information (24% vs. 33%), the substitution rate for Google is 95%, and almost everyone considers AI to be more effective, even when it comes to preparing a purchase. These two French trends are not set in stone. The penetration rate of AI in the younger generations (82% among 18-24 year-olds), the sustained frequency of use by adopters, and the fact that one French person in five has already substantially reduced his or her reliance on Google all point to a clear trajectory. The proportion of regular users in the population will mechanically grow, and with it, the place of AI in search paths, including commercial ones. For brands, the challenge is to anticipate this shift. The behaviors observed today among early adopters – searching for information via AI, comparing products by conversation, expecting personalized recommendations – foreshadow the expectations of the median consumer in a few years’ time. Those who have invested their visibility in AI responses will be in a position of strength; those who have waited for proof of the majority shift will have to make up for lost time.
