Ipsos study for Eskimoz: French shoppers turn to AI as Google loses ground

On 28 October, the French tech publication Siècle Digital reported on a study carried out by Ipsos for Eskimoz on how French consumers use AI, published as the Christmas shopping season approaches. The headline figure: 52% of French people now use an AI tool at least once a week. The article's framing is that AI is taking a growing share of purchase research and that Google is losing ground, which is different from saying Google has been replaced. The detailed breakdown of the study by product category, age group or stage of the purchase is not covered here.

One number is enough to change how a marketing team should think about the end of the year. A tool used weekly by half the population is no longer an experiment; it is part of the routine people bring to a purchase decision.

A weekly habit reshapes the path to purchase

As general context, conversational assistants are used in shopping in a few recurring ways: to narrow a broad need into a shortlist, to compare two or three products, to check a technical detail, or to draft a gift idea for someone else. Each of these steps used to involve several searches and several site visits. With an assistant, many of them happen in a single conversation, and the user may only visit a site at the very end.

For acquisition, that has two consequences. First, the first contact with a brand is increasingly a mention inside an answer rather than a visit. If the product is not named or described accurately at that stage, the brand is not in the shortlist, and no amount of retargeting will bring back someone who never arrived. Second, the visits that do come from assistants tend to arrive later in the decision. They are fewer, but the user has already done the comparison work.

For monetisation, it means landing pages and product pages receive a visitor who expects confirmation, not explanation. Someone who arrives after an assistant has recommended a product wants to verify the price, the delivery date and the return policy, then buy. A page built to educate a cold visitor slows that person down.

What it changes before the holiday peak

Google remains the main acquisition channel for most retailers, and "losing ground" describes erosion, not replacement. The practical question is how to protect the AI-mediated share of demand without moving budget away from what already works. Most of the work is on product information, not on media spend: assistants can only recommend what they can read, and they tend to pick up structured, consistent, complete information about price, availability, specifications and policies.

Reviews and third-party mentions also matter, because assistants draw on what others say about a product, not only on what the brand says about itself. A product with thin or outdated information in public sources is harder to recommend.

What to check this week

• Ask the main assistants the questions your customers ask before buying in your category, and note whether your products appear, and whether price and characteristics are described correctly.
• Check that every key product page shows price, stock, delivery date and return conditions without scrolling, and that the same data is in your structured product markup and merchant feeds.
• Segment AI assistant referrers in your analytics so you can compare their conversion rate and basket value with organic search and paid traffic.
• Review your top holiday landing pages from the point of view of a visitor who has already decided: remove anything that delays the add-to-cart.
• Look at where your products are reviewed or discussed outside your own site, and fill gaps before the peak rather than after.

The 52% figure does not tell you how much of your revenue already passes through an assistant. Your own referrer data and a few manual checks will tell you more, and both can be done before the season starts in earnest.