Does AI-generated traffic attract more customers?| Firstcom

Understanding GEO

A new way of searching the web

For a long time, online visibility meant one thing: appearing at the top of Google’s search results. Search Engine Optimisation (SEO) primarily focused on ranking website pages among the top results in search engines, maximising visibility and attracting visitors.

With generative AI, the user journey is changing. Instead of typing three keywords into a search engine, users turn to an AI assistant such as ChatGPT or Gemini. They ask a full question, specify their budget and compare options. Search is becoming a conversation. Users refine their queries, ask follow-up questions and seek recommendations. The assistant responds with a summary that names companies and provides links to their websites.

And this shift is no longer limited to ChatGPT for French internet users. Since 22 July 2026, Google has also been displaying AI-generated responses in France through AI Overviews, alongside a conversational AI Mode directly within the familiar search engine. Millions of French internet users are therefore getting used to receiving written answers rather than a list of links. All signs suggest that this trend is here to stay and will continue to grow.

How do AI assistants mention brands?

To generate an answer, an AI assistant draws on what it has learned and, when available, information retrieved from online searches at the time of the query. It selects a small number of pages and uses their information to build its response. Brands can appear in different ways. Some are actively recommended, with a direct link to the relevant page. Others are simply mentioned by name, without a link. Some appear only in the sources cited, well away from the actual recommendation.

These three situations do not have the same value. Being read by AI is not enough: to generate visits, a brand needs to be mentioned and clickable when users are making their decision. This is precisely where GEO comes in. GEO (Generative Engine Optimisation) is the practice of optimising a brand’s online presence so that it is mentioned and recommended in generative AI responses.

What this means for SEO

SEO is not disappearing: it is becoming the foundation. A poorly structured page that is difficult to crawl is unlikely to be picked up by an AI assistant. But ranking well is no longer enough. Your content must also clearly answer the questions users ask AI assistants, with accurate, up-to-date and verifiable information: product features, prices, service areas, comparisons and answers to common objections.

On the technical side, AI systems must actually be able to read your pages. Their crawlers need access, content must be directly readable in the source code, and Schema.org structured data needs to be properly implemented. Without a solid technical GEO foundation, even the best content can remain invisible. To find out more, read our article on technical GEO optimisation: why AI systems can’t see your website and how to fix it.

GEO builds on SEO. The goal is no longer simply to rank, but to be selected as the answer.

What is AI-driven traffic really worth?

When users click through from a conversational AI platform, they have often already completed much of their decision-making journey. They have explained what they need, ruled out certain options and read an initial overview of your offering. On paper, they look like potential buyers. But we wanted to verify that with actual data.

We therefore analysed website data from some of our clients for whom we provide GEO services. We looked at two different business models: e-commerce websites and websites where conversions take the form of quote requests or contact enquiries.

For each, we compared traffic from AI assistants with traffic from search engines and paid campaigns over a 90-day period.

E-commerce: visitors who actually buy

Our first example is an e-commerce business specialising in consumer electronics.

Over the past 90 days, AI assistants accounted for just over 1% of its website visits. Yet these visitors were 1.7 times more likely to make a purchase than those arriving from Google, and 2.7 times more likely than visitors from paid campaigns. They also added products to their baskets almost twice as often as Google visitors.

Their average basket value was slightly lower, but they arrived with a much clearer idea of what they wanted. The result? Each visit from an AI assistant generated almost 1.5 times more revenue than a Google visit, and more than twice as much as a visit from paid campaigns.

Fewer visits, but more buyers.

Leads and quote requests: where the difference becomes even more striking
For websites focused on quote requests, the findings are even more compelling.

On the website of an automotive retail network, visitors coming from AI assistants submitted almost twice as many quote requests as those arriving from Google, and 2.6 times as many as visitors from paid campaigns.

Their bounce rate was 2.4 times lower than that of Google visitors and four times lower than that of paid campaign visitors. They also spent longer on the website, performing nearly seven actions per visit, compared with five for Google traffic.

These are not casual browsers: they arrive ready to explore a project that is already underway.

The same trend can be seen at a B2C company with around 100 locations across France. AI-driven traffic generated three times as many quote requests as Google traffic and 3.4 times as many as paid campaign traffic.

Interestingly, these visitors were less likely to call a local outlet directly. Instead, they tended to submit a quote request through the website.

Finally, consider an independent tradesperson offering a highly technical, premium, made-to-measure product. AI accounts for only around 4% of website visits, but generates nearly 12% of contact enquiries. Visitors from AI assistants are twice as likely to request a quote as those arriving from search engines.

For this type of project, the figures make perfect sense. Before contacting a tradesperson, buyers have countless questions about how the product works, installation, budget and lead times.

AI answers many of these questions beforehand, effectively acting as an initial adviser. By the time visitors click through, they are better informed, and twice as many take the next step and get in touch to discuss their project.

Our findings at a glance

Client	Conversion measured	AI traffic compared with search engine traffic (Google)
Consumer electronics e-commerce business	Orders	1.7 times higher
Automotive retail network website	Quote requests	Almost 2 times higher
B2C company with around 100 locations across France	Quote requests	3 times higher
Premium made-to-measure product specialist	Quote requests	2 times higher

Why do these visitors convert better?

Because they arrive later in the decision-making process.

Before clicking through, they have already asked their question, received an answer and compared different options. The conversational AI assistant has already done some of the work that websites traditionally had to do themselves: filtering options, explaining products and services, and building confidence.

By the time visitors arrive on your website, they are no longer discovering the subject. They are there to verify what they have learned and take action.

This is our interpretation, and the figures we analysed all point in the same direction: AI-driven traffic converts between 1.7 and 3 times more often than traffic from other acquisition channels.

In other words, GEO is not primarily about attracting more visitors. It is about attracting better-qualified visitors: people who are already informed, have compared their options and are ready to take the next step.

GEO is already measurable

Every visit recorded under the AI Assistants channel tells a similar story: someone asked a question, an AI assistant highlighted a link to a brand, and the user chose to click through.

One important point
Some AI responses mention or recommend a brand without providing a direct link. Although this makes direct attribution more difficult, many users subsequently search for the brand by name in a traditional search engine to continue their journey.

This indirect traffic generates additional, tangible value for businesses investing in GEO (Generative Engine Optimisation), even if it is harder to attribute than a direct click from an LLM response.

For the clients we work with, this is no longer just a hypothesis. Their brands appear in AI assistant responses, those responses generate website visits, and those visits convert.

One thing is clear: AI assistants do not mention brands at random. They draw on brands whose authority and content provide clear, precise and verifiable answers to users’ questions.

For businesses, the challenge is no longer simply to be found, but to be recommended. And that is something you can work on.

Want your brand to appear in AI-generated answers?

Through GEO, we optimise your content, authority and website so that ChatGPT, Perplexity, Gemini and Google’s AI Overviews mention and recommend your brand. Our clients are already seeing measurable results.

KEY TAKEAWAYS

ChatGPT, Perplexity and Gemini still send relatively little traffic to websites. But these visitors behave differently from others. We analysed 90 days of Matomo analytics data from some of our clients to measure the actual value of this traffic. The trend is clear: arriving later in the buying journey, these visitors convert 1.7 to 3 times more often than those coming from search engines or paid campaigns.