Why Japanese E-Commerce Giant Rakuten Is Betting Big on AI

“We need a revolution,” said Japanese e-commerce giant Rakuten’s head of fashion Ryo Matsumura. “Brands need to express their image and identity to consumers, but if the digital solution is limited to video, photo and text, that’s very limiting. [What they really need are] more cutting-edge technologies to express themselves.”

To that end, Rakuten has released a suite of AI tools providing brand merchants with what’s called an NLUI — a ‘natural-language user interface’ that works with written or spoken colloquial commands instead of IT jargon — to harness data insights across its platforms. Launched in 2023, the backend tool was just the start. To make the transition to digital more intuitive for Japanese consumers, many of whom still eschew online shopping, the company has been investing in generative AI search and discovery.

The end-game, of course, is to leverage AI to boost sales. Last year, Rakuten Fashion, the group’s dedicated portal for designer labels, launched semantic search — a technology that applies deep learning AI to interpret the meaning of search terms rather than taking them at face value or relying on keywords.

The company found that semantic search resulted in a 93.5 percent reduction in zero-hit-rate scenarios. Through the technology, Rakuten has been able “to increase search per [user] session by three percent; increase sessions per user by four percent; and therefore increase overall sales through search by five percent,” Rakuten Group chief AI and data officer Ting Cai said in an interview late last year on the company’s YouTube channel.

Rakuten has since rolled out the feature for users of its main marketplace Rakuten Ichiba, and plans on doing the same for its online flea market Rakuten Rakuma and Rakuten Books next year.

There’s a reason fashion was the testing ground for the new search function. Rakuten’s other major e-commerce categories — fast moving consumer goods like food and beverage items, and electronics — provide a much more straightforward roadmap when it comes to enhancing online consumer experience.

In fashion, where personal style and taste hold more weight and virtual experiences must hold a candle to the country’s commitment to brick-and-mortar hospitality, tech players are up against tougher competition. Alongside investing in 360° product displays and virtual try-on tech, Rakuten is also banking on AI usage across product descriptions, images and sizing to make the online purchasing experience more akin to an offline shopping experience.

“We want to leverage AI to deliver the ultimate hospitality to our customers. Not only to meet their needs, but to anticipate what they want,” Cai said at an annual Rakuten event this year, referencing the Japanese concept of omotenashi, or selfless service. It is that unparalleled in-store service that — alongside highly favourable exchange rates — propelled international tourists to Japan last year, fuelling a local luxury boom.

In recent years, Rakuten has been aggressively pursuing luxury brand partnerships in a bid to grow its fashion segment and to counter the efforts of competitors.

Japanese rivals include Zozo, whose online fashion Zozotown platform has been expanding its luxury brand offering on its dedicated fashion portal Zozovilla. Moreover, most of the top department groups selling luxury goods in the country — including Isetan Mitsukoshi, Daimaru Matsuzakaya, Takashimaya and Hankyu Hanshin — have ramped up efforts to digitise since the pandemic.

AI will continue to be a key battleground for these and other e-tailers in the years ahead, both in Japan and overseas.

Damien Yeo, Fitch Solutions’ Asia-Pacific consumer industry analyst, notes that other Asian e-commerce players, including JD.com and Alibaba’s Taobao in China, are already using AI algorithms to analyse customer behaviour to tailor product recommendations, alongside AI-powered chat bots. There is also growing AI usage for trend prediction, supply chain optimisation and visual search.

At this point in the AI race, Yeo says, it’s hard to accurately size up Rakuten against its peers in terms of return on investment. But one of the areas where he believes Rakuten has a competitive edge (both within and beyond Japan) is its vast data depository, which is rich in insights into business and e-commerce trends.

“The group has more than 1.8 billion members interacting with over 70 service offerings by the company…Ultimately, Rakuten’s USP at the moment lies in its wide coverage of various retail segments and [its ability] to leverage the Rakuten platform to encourage cross-sales from other retail segments to its fashion business,” Yeo explains.

However, Yeo contends that where Rakuten’s role in Japan makes it a jack-of-all-trades akin to Amazon, Zozo is still considered the local market leader for fashion e-commerce.

Rakuten’s Matsumura, who also serves as group managing executive officer, declined to disclose the amount the company is investing in innovations like AI, but if his billionaire-boss’ stated ambitions are any indication, it must be a sizable figure.

“AI is going to change everything [and] what the Rakuten Group wants to do is democratize AI,” said company founder and chief executive Hiroshi “Mickey” Mikitani at an August company conference. “[That’s] more than just a vague suggestion to use AI tools… We’re setting [extremely bold and] concrete targets to utilise AI.”

The company initiative Mikitani described is titled “Triple 20,” which if successful would see the firm boost “marketing efficiency by 20 percent, operational efficiency by 20 percent, and the efficiency of our clients — such as the merchants and hotels on our platforms — by 20 percent,” in Mikitani’s words.

In addition to adopting external tools like generative AI and large language models (LLM) from OpenAI, Rakuten has released in-house tools such as Rakuten AI 7B, a set of LLM that excels in the Japanese language and works at a smaller scale to the likes of OpenAI’s ChatGPT, Mikitani said.

If Rakuten’s innovation efforts go to plan, they will dovetail with a broader shift in the local fashion industry, which is seeing higher-end designers, and eventually hard luxury players, make online transactions a higher priority.

“Some designer brands are thinking about [digital] sales channels more seriously, and eventually higher-priced luxury products will follow suit,” said Matsumura, who oversees relationships with Rakuten Fashion’s existing partner brands which include Maison Margiela, Marc Jacobs, Sergio Rossi and Chloé, alongside niche labels like JW Anderson and Anrealage, and private label collections from multi-brand Japanese retailers Beams and United Arrows.

The pandemic was seen as an opportunity to nudge local luxury shoppers toward digital channels in the hope of mirroring the online shopping boom in markets like China. While Japan — Asia’s most mature luxury market — is still lagging China’s enthusiasm for e-commerce, its consumers are increasing uptake of it.

Japan’s online sales across all personal luxury goods accounted for just nine percent of all global sales in 2019, says Fflur Roberts, head of luxury goods at Euromonitor International. That figure rose to 13 percent in 2024 when online sales of the designer fashion category made up 17 percent of overall sales, up nine percent from pre-pandemic levels, and just one percent behind neighbouring South Korea.

These are figures Rakuten is keen to see grow further.

“The change has just started,” predicts Matsumura, noting the continued dominance of physical retail, which despite Japan’s shrinking department store network, remains integral to the local shopping culture. Indeed, the maturity of Japan’s luxury industry is what’s holding it back from more fully embracing the online shift.

“Obviously the sale structure has been changing since the pandemic, [but] the percentage of digital transactions should grow over the next three-to-five years,” Matsumura asserts.

AI is primed to speed up this transition.

According to Euromonitor’s Roberts, gamification efforts and technologies like generative AI will be key in markets including Japan and South Korea as a growing number of sales take place through digital devices, and shoppers shift from viewing e-commerce as a means to an end to an enjoyable experience in itself.

However, the shift won’t be straightforward for Japanese businesses — or for global brands looking to target local online shoppers.

Indeed, there are potentially difficult economic and political headwinds for Japan to reckon with first. With choppy waters ahead, Matsumura is betting on Rakuten’s integrated banking business — which includes an in-house credit card and points system — to add value to its various marketplaces.

But company performance hasn’t been entirely positive. In the group’s Q2 results published in August, Rakuten reported that domestic e-commerce sales across its internet services segment (spanning Rakuten Ichiba as well as specialised platforms such as Rakuten Fashion and Rakuten Beauty) were down 4.8 percent year-on-year — a decline the company attributed to the tapering of a government-subsided nationwide travel initiative and revisions to its points program.

The goal is to return to growth from the fourth quarter onwards, while continuing to roll out new tech across the Rakuten umbrella. It remains to be seen whether AI investments will break the ice and endear more luxury and premium brands to Japanese marketplace players as they did in China on platforms like Tmall and JD.com.