Luxury Briefing: How LVMH is tackling luxury returns with AI

In December, Amazon filed a complaint against REKK, a group with members based in the U.S and Europe who post tips on Reddit about how to fraudulently return items to a range of stores. Among them: Sephora, Nike and Walmart. Groups like these suggest that you can get a refund for returning an empty box or get refunds of up to $300 based on a lie about an item not arriving at the destination address.

However, when brands have attempted to combat the issue by turning to customer tracking services like risk score company Retail Equation, they have been faced with an outcry from customers over system failures. 

The problem only seems to be getting worse. According to a January 22 report from returns solutions company Loop Returns, 39% of U.S.-based online shoppers admit to having either engaged in returns policy abuse themselves or know of someone who has. Between 20-30% of these respondents have engaged in returns policy abuse behaviors at least once per week. Respondents’ reasons included needing an item just for a specific event (36%), taking advantage of a lenient return policy (23%) or wanting to keep the item without paying for it (15%).

Brands’ solutions have included punishing customers and working to maintain a firm returns policy.

Like other sectors, return fraud, including “wardrobing,” is a headache for luxury brands. For example, LVMH is reporting a rise in fraudulent return rates.

“Currently, depending on which brand it is, the return rates [for LVMH brands] are typically between 20-30%,” said Anders Lange, LVMH in-house Committee member, who guides the strategic retail direction for Louis Vuitton, Christian Dior, Givenchy and Sephora, among others. “And for every return on that, there’s 14-15% return fraud. If you look at the data on those customers, the profile is someone that keeps buying and returning, or even tries to return fakes instead of the real merchandise. … And it isn’t Gen Z doing it. It’s those on an average income of around $50,000; it’s the group that has the money to buy luxury occasionally.”

He added, “As there is a growing interest in influencers, we are also seeing wardrobing increasing. They’re buying it and then returning it after taking an Instagram picture or wearing it for a party.”

Wardrobing is a form of return fraud where a customer purchases an item, such as clothing or electronics, uses it for a short period, and then returns it for a full refund. This practice is particularly common with items intended for one-time use, like a dress or handbag for a wedding or other special occasion.

To prevent this, brands have tried affixing big tags to clothes that can’t be removed for the item to be returnable, said Brooke Bulan, COO at luxury advisory company Fixer Advisory. But, in doing so, they often transfer the costs associated with fraudulent returns to their own handling process.

As reported on Thursday, LVMH’s sales in the fourth quarter increased by 10% year-over-year, surpassing analyst predictions. In its fashion and leather goods division, which is the largest in terms of revenue and profit, there was a 9% rise in sales, reaching €11.3 billion, in line with expectations.

Sarah Davis, co-founder and president of resale marketplace Fashionphile, said that, for resale companies, return fraud is often a bait-and-switch return. “This is where a criminal purchases an authentic item and then exchanges that item with a very close counterfeit of the same style,” said Davis. And “superfakes” of luxury items, including bags, have become more prevalent.

“Although all luxury brands have return policies, they also have these very high-touch relationships with clients, and there’s a lot of sensitivities there,” said Bulan. “Sometimes excessive returns becomes much more of a relationship management issue than an actual policy enforcement.”

Brands blacklist customers, but then those shoppers often move on to other brands, Bulan said. “It doesn’t change their mental state or the issues around return fraud.”

Currently, more brands are seeing potential in using AI to predict which customers are going to commit return fraud or have a high return rate, she said.

For its part, LVMH is doing tests of a predictive returns AI that launched six months ago. It aims to bring down returns by around 50%. The predictive AI will be able to analyze store associate data and returns information and see which customers return the most items across in-store and e-commerce. 

“We are starting to create a return policy that fits the individual customer,” said Lange. “That means: If you exhibit a certain behavior, the return policy changes specifically for you. If you have a high return rate, you might get a shorter [return window], or you get a fee on it.”

He added, “[We want to] minimize how long customers have to return the goods and [require them to] bring the returns down to the store,” he said. “Looking a person in the eyes and saying, ‘I don’t want it,’ is harder than just putting it in a bag and sending it. And it is a matter of investment for the consumer because they have to invest the time to go down to the store, too.”

LVMH’s other solutions being planned include a “returns staircase,” said Lange. After seven days, you get a full refund; after 14 days, you get a 90% refund; etcetera. “Because, technically, [keeping items for a long period] means you’re renting it. But you’re still not allowed to take the labels off [to return it],” he said.

According to Nicole Marra, founder and CEO of Fixer Advisory, brands have considered finding alternate revenue streams for products that may have been “wardrobed” by offering them in resale or rental channels. This can alleviate the risk of confronting a customer about a questionable return that may be legitimate.

Using third-party returns services like Loop is another way brands are aiming to alleviate returns-related losses. These companies deal with large amounts of customer data and return logistics, which can help identify patterns of suspicious activity and help stop returns fraud. For its part, Loop is developing a predictive AI tool, similar to the one being trialed by LVMH, which is set to be implemented by the company later this year. Loop’s clients include luxury brands Erdem, Khaite and Gabriela Hearst.

“More and more brands are going to be looking at infrastructure that they can put into place, whether it’s through training or physical tools or working with third-party vendors who have expertise in this area,” said Marra. “This will allow them to keep a good customer experience while still reducing some of this fraud and making it harder without explicitly saying so.”