An assistant offers to “find your ideal size” in just a few questions. On Instagram, a video shows the exact same outfit on an influencer you follow. Within minutes, the decision almost makes itself. This now-commonplace scenario illustrates the growing role of artificial intelligence in e-commerce.
French e-commerce reached €196.4 billion in revenue in 2025, up 7%, with 3.2 billion transactions and 42.2 million online shoppers, according to Fevad. This growth comes with a deeper transformation: platforms use browsing data, purchase histories and preferences to personalize the experience in real time. This evolution of e-commerce is part of a broader digital transformation driven by innovations in artificial intelligence and regularly analyzed by specialized media such as Silicon Valley.
Algorithms no longer simply rank search results. They anticipate, suggest and, increasingly, hold conversations. Nearly one in three online shoppers already uses generative AI in their purchasing journey, mainly to look up information, compare products or prepare their decision, according to Fevad’s 2026 key figures. On the business side, 94% of e-merchants use generative AI solutions.
![]()
Artificial intelligence is entering our shopping habits
Retail sites continuously analyze behavior: pages visited, time spent, items added to the cart, returns, reviews consulted. These signals feed models that adapt the homepage, the suggestions and sometimes even the order in which products are displayed. On large general-purpose platforms as well as on specialized fashion or beauty sites, the experience becomes individual. The same catalog no longer looks the same to two different visitors.
This personalization relies on collaborative filtering techniques (comparing your tastes with those of similar profiles) and content-based filtering (analyzing the characteristics of products you have liked). The data is cross-referenced with context: season, approximate location, device used. The result is visible: fewer manual searches, more suggestions that “match you.” According to an Odoxa-Fevad study, 31% of French online shoppers already use generative AI for their online purchases. Among regular users of conversational tools, that proportion rises sharply.
The limits quickly become apparent. Recommendations can lock people into a bubble of already-known tastes, reducing chance discovery. They also encourage impulse purchases when the offer arrives at the precise moment the desire forms. Platforms are aware of this: the commercial goal remains to increase the conversion rate and the average basket.
![]()
Increasingly personalized product recommendations
Recommendation systems are at the heart of the experience. Amazon has long illustrated their power; other players have followed. Today, they rely on more sophisticated models, sometimes generative ones, capable of producing explanations or comparisons. An algorithm may suggest a wool coat because you bought warm sweaters the previous winter, browsed neutral shades and spent time on ankle boot pages.
In fashion, suggestions take into account style, dominant colors and sometimes the body shape you have declared. In beauty, they factor in skin type, products already ordered and searches (“long-wear foundation”). In home decor, they propose objects consistent with a Scandinavian or bohemian style identified from your favorites.
The benefits are concrete: time saved, fewer doubts about relevance, better fit. Market studies and platform feedback show that sessions with interactions on recommendations often show a higher basket value. But there are drawbacks. Being locked into past preferences limits exploration. Overly insistent recommendations, fueled by data that is sometimes collected opaquely, can influence the decision more than the actual need. Telling an objective algorithmic suggestion apart from a commercial promotion remains difficult for the consumer.
Virtual assistants and chatbots make shopping easier
Chatbots and conversational assistants have evolved. They no longer just point to an FAQ. They compare products, answer precise questions about materials, dimensions or composition, and suggest gift ideas based on a budget and a recipient. Amazon recently rolled out “Alexa for Shopping” in France, which draws on history, preferences and product data to prepare a list, compare items or track price changes. The platform claims that more than 400 million customers have used its AI shopping assistance tools over the past twelve months.
![]()
In fashion and beauty, these assistants point toward complementary pieces or skincare routines. In home decor, they help visualize color combinations or suggest accessories that suit a piece of furniture already chosen. The customer experience improves: fewer abandoned carts due to lack of information, immediate answers, sometimes available 24 hours a day.
Since August 2, 2026, Article 50 of the European AI Act has imposed a transparency obligation: chatbots and assistants must clearly state that they are artificial intelligence systems from the very first interaction. This rule, clarified by the European Commission’s guidelines, aims to prevent any confusion between a human exchange and an automated one.
Fashion and beauty: AI helps choose the right products
These two worlds concentrate particularly visible applications. Clothing recommendations based on style preferences are becoming more refined. Systems cross-reference purchase history, returns (what didn’t work) and browsing to suggest better cuts and sizes. Zalando has built size advice directly into its conversational assistant, drawing on size guides, the customer’s history and reviews about the actual fit. The retailer has also developed virtual try-ons: from a few photos, an avatar lets you see how a garment hangs and helps reduce size-related returns.
Visual search is advancing: photograph a garment seen on the street or on a social network to find similar models on a platform. Makeup simulations, powered by augmented reality technologies such as those from ModiFace (acquired by L’Oréal) or Perfect Corp, let you virtually try on a lipstick or foundation on your face via the camera. These tools reduce the uncertainty of buying remotely and encourage experimenting with shades you wouldn’t have dared to try in a store.
![]()
The limits remain real: the accuracy of a virtual try-on depends on the quality of the photos and 3D models. A size recommendation is still an estimate. Consumers are still better off checking reviews and the measurements provided.
Decor and lifestyle: imagining before buying
Visualizing a piece of furniture or an object in your own home is one of the most concrete uses of AI and augmented reality. IKEA popularized this approach with its Place app, which overlays true-to-scale 3D models through the smartphone camera, and then with IKEA Kreativ, which lets you scan a room, virtually remove the existing furniture and test complete layouts. These tools reduce sizing errors and help evaluate color and style harmony before ordering.
Other platforms offer decor inspiration generators: from a photo or a description (“cozy living room, warm tones, small budget”), the AI suggests combinations of furniture, textiles and accessories, sometimes with direct links to available products. These features make it easier to project yourself into a space and reduce returns caused by a gap between the online image and the reality of the room.
![]()
Social media and AI shape the urge to buy
TikTok, Instagram and Pinterest play a growing role in discovery. Their algorithms analyze watch time, likes, saves and interactions to push personalized content. A beauty routine video or a fashion look can appear just when the user is most receptive. Social commerce is growing: according to an Ifop barometer for Colissimo, 15% of French people say they have already made a purchase directly on a social network in 2026, up from 9% the previous year. Among those under 25, the figure reaches 39%.
Sponsored content, creator collaborations and viral trends blend with algorithmic recommendations. Telling organic discovery apart from paid promotion sometimes becomes tricky. Pinterest, more geared toward intentional search, remains a favored tool for decor and lifestyle, where users plan more than they buy on impulse. TikTok, more impulsive, favors quick discoveries and direct purchases through TikTok Shop in markets where the service is available.
Personal data, transparency and impulse buying
This personalization raises concrete questions. The collection and use of browsing, purchase and preference data make consumer profiling possible. The GDPR regulates this processing: an appropriate legal basis, clear information, and rights of access, rectification and objection. Cookies and trackers used for commercial personalization and for training recommendation algorithms generally require prior consent. The CNIL recently reminded the market of this by fining Fnac Darty a total of €7 million (October 2026) for using trackers in its mobile app that collected browsing history to display product recommendations and train its algorithm, without valid consent.
The European Digital Services Act (DSA) imposes transparency obligations on the recommendation systems of very large online platforms. Users must be able to understand the main parameters and, in some cases, modify or disable personalization. The lack of transparency of some algorithms, the sometimes overly insistent commercial recommendations and the difficulty of distinguishing objective suggestions from sponsored content persist. Impulse purchases, encouraged by well-calibrated suggestions and smooth interfaces, are a real risk, especially when the budget is already tight.
Data protection issues remain central. Platforms must document their processing, limit data to what is necessary for the stated purposes and enable the effective exercise of rights. Consumers are well advised to read privacy policies and personalization settings.
How can you use AI to buy better online?
These tools can make shopping more convenient and better tailored, provided you keep your freedom of choice. A few simple habits help you benefit from them without letting yourself be entirely guided.
Comparing several products and several stores before confirming an order remains essential. Recommendations are suggestions, not final decisions. Checking detailed specifications, dimensions, composition and recent reviews limits unpleasant surprises. Comparing prices and delivery costs across several sites avoids paying more for a similar item. Paying attention to sponsored content and “recommended for you” labels helps you better identify where a suggestion comes from.
![]()
Reading privacy policies and cookie or personalization settings gives you better control over the data you share. Avoiding unnecessarily sharing precise personal information with virtual assistants reduces risks. Finally, taking time to think, especially when faced with a highly targeted suggestion or a limited-time offer, helps you tell a real need from an impulse.
Artificial intelligence no longer merely makes it easier to find products. It now takes part in how consumers discover, compare and select what they buy. It can make the experience smoother and more personalized, provided everyone keeps their free will and stays attentive to how their personal data is used. The tools are evolving quickly, and so are habits. Understanding how they work makes it possible to choose better, rather than letting yourself be chosen.