In recent years, several websites selling ultra-low-cost goods have appeared on the French market. Shein, Temu and AliExpress, to name but a few, are shaking up the online retail landscape. According to a study conducted by BPCE Digital & Payments, the number of payment cards recording at least one monthly transaction on a discount site increased by 20% between the first quarters of 2022 and 2023.
This is hardly surprising given that Temu’s website has millions of French visitors every month, according to data from the Federation of E-commerce and Distance-Selling (FEVAD). In mid-July 2025, low-cost platforms accounted for 22% of parcels handled by the French postal service, compared with 5% five years ago. This increase is set to continue, with the sector expected to grow by a total of 6.5% in 2025.
Of course, the rampant inflation in France in recent years partly explains this craze. But this is not the only explanation for these developments. The use of artificial intelligence (AI), which is at the heart of the business model of these platforms, helps to build consumer loyalty.
Behavioural profiling
In our latest articles on Shein and Temu, we analysed how these platforms operate behind the scenes. By analysing user behaviour data, the AI tools used by the platforms can identify the customers most likely to make a purchase and tailor the advertising messages they receive.
Predictive algorithms also analyse user behaviour to offer personalised recommendations. This approach aims to create a need before it even arises, by playing on feelings of scarcity and urgency. This is the famous Fomo, or “fear of missing out”.
These predictive algorithms have been around for many years, but their new capabilities, “augmented” by AI tools, are ushering in a new era, adapting even more finely and quickly to each Internet user. At the bottom of each page, there is a list of “items also viewed” by other users, which are similar to the product being searched for. This classic marketing technique is taken a step further: algorithms constantly submit new content to the customer to study their reaction. Every reaction (such as a click to add an item to a basket) is analysed in real time. The algorithm, supported by AI, then uses this data to encourage the user to buy other products that they did not initially come to look for.
Using play to sell
Gamification refers to the use of game mechanics for marketing purposes to capture customers’ attention.
On the Temu app, the interfaces are inspired by gambling games, which are known to be particularly addictive: wheels of fortune, countdowns highlighting limited-time offers, gifts and promotional codes to unlock, etc. These constant stimuli create a sense of urgency in the user, while disrupting the biochemical mechanism of the reward circuit. AI allows for greater precision and diversification of the “games” offered, to the detriment of buyers (and without their full awareness).
On Temu, mini-games integrated into the mobile app (such as Farmland and Fishland) promise free items and discount coupons. Points and voucher systems are used to encourage users to return to the site as often as possible, and personalised notifications are also sent at opportune moments, based on data collected about the user.
In addition, dynamic pricing algorithms (which adjust prices according to variations in demand) display discounts that can have a powerful effect on consumers. Here too, AI proves to be a tool that multiplies this power.
Personalised online stores
The hyper-personalisation of platforms is another lever. Thanks to AI, which collects vast amounts of data on user profiles, each customer has a different online store, personalised according to their history, tastes, preferences and aversions. This increases the likelihood of one or more impulse purchases.
AI’s most significant contribution to Shein’s success goes much further, and precedes the arrival of customers on the platform. Shein has developed its own AI tools and algorithms to collect and analyse data. Using these tools to track customer behaviour on the Internet (on and beyond its website), Shein also relies on them to analyse online search results, social media posts, and competitors’ websites.
These tools are central to Shein’s success. They enable it to identify trends (colours, prices, designs) in real time or near real time, and to adjust the design and production of its products very quickly. That’s because all this data is shared with suppliers who produce the items Shein sells. The process is facilitated by a strategy that favours small-volume production (100 items or less) for all new products.
Significant ethical issues
All of these factors raise ethical issues, given the opacity of the algorithms and the lack of transparency regarding the use of the data collected.
In 2022, Shein was fined by the state of New York for failing to inform nearly 40 million users of a data breach that occurred in 2018. As mentioned above, the company is also being investigated by the European Commission, which accuses it of at least six deceptive or abusive practices toward consumers (such as “fake discounts”, “misleading information”, “hidden contact details”, and others).
So, to what extent should AI be regulated in online sales and marketing? What limits should be set? How far should consumer protection go? According to a 2024 Statista report, AI-based recommendation systems influence nearly 35% of online purchases, demonstrating their considerable impact.
The EU’s Digital Services Act (DSA) and its AI Act both strengthen consumer protection by improving transparency, accountability and safety in the digital sphere. The DSA aims to ensure that users are protected from illegal or harmful online content, manipulative designs and opaque algorithms. The AI Act prevents the use of high-risk or deceptive AI systems, requires clear disclosure when AI is involved, and enforces safeguards against bias, discrimination and abuse. But there are questions about scope and impact: how do these regulations specifically address the risk of opaque or questionable recommendation algorithms? What mechanisms will be put in place to ensure compliance and enforcement of digital actors? Ultimately, how can we measure the regulations’ effect on consumers and their protection?
- is an Associate Professor of Innovation and Strategy, EDHEC Business School
- is a Director of Pedagogy – Full Professor, IÉSEG School of Management
- This article first appeared in The Conversation
