Games people play: How Shopee uses data science to make shopping a holistic experience

Online shopping in Southeast Asia has truly taken off, and data shows it will get even bigger.

According to the e-Conomy SEA 2019 report released by Google, Temasek, and Bain & Company, 49 million people in the region either bought or sold items online in 2015. Just four years later, that number tripled to 150 million.

With the rise in users comes an increase in money spent. The ecommerce market was worth US$32 billion in 2019, making it the biggest sector in the internet economy, ahead of other heavyweight industries like online travel and ride-hailing.

What are ecommerce firms doing to stand out in such a highly competitive field?

As the attention span of users gets shorter, ecommerce players need to come up with different ways to keep them engaged, says Pan Yaozhang, head of data science at Shopee. “This can include games, livestreaming videos, and other interactive features.”

Online shopping is often done using mobile phones, and this is especially true for Shopee, where over 95% of orders are made on its mobile app, adds Pan. “Our platform was built to be highly mobile and social from the start,” she says.

Pan Yaozhang, head of data science at Shopee / Photo credit: Shopee

To retain users, the company has made shopping a more holistic experience by applying data science to focus on three key areas: personalization, engagement, and social.

Personalization

Southeast Asian consumers are diverse. That’s why Shopee employs artificial intelligence and big data to interpret large data sets that allow it to identify “relationships and patterns in users’ shopping behavior,” shares Pan.

To analyze the data and build AI models that profile the company’s entire user base, Shopee’s data science team works with its local marketing and design counterparts. Localization is important, as the region generates such a huge amount of data that it’s “difficult to generalize to all markets,” says Pan. Data science helps the team make sense of local data and build models that are able to carry out functions like specifying which customized banners and marketing assets are displayed to each specific user group across the markets.

Diving deeper into the shopping experience, data science also plays a part in building shopper models for each market, based on various behavioral and demographic data such as gender, age, brand preferences, and preferred shopping features.

“In Indonesia, for example, we find from the data that users can be grouped into different profiles based on what brands and categories they prefer,” observes Wang Chen, data science lead for marketing science. “This allows us to make the shopping experience more intuitive, with fine-tuned recommendations.”

Meanwhile, users in Singapore have a strong preference for flash sales and limited-time promotions. Shopee taps into this by diving “deep into the data to discover what products and categories are popular across the day,” explains Wang.

“This allows us to better plan our flash sales calendar and curate a better shopping experience for users every day.”

Engagement

While buying things online might seem like a quick and mundane process, Shopee works toward engaging its customers more and keeping them on the app for longer.

Shopee Games does exactly this. Users can play games in exchange for rewards such as Shopee coins – the platform’s virtual currency that can be applied at checkout to get discounts – or for brand-sponsored products.

Data science has also been instrumental in game development. An observation that users are actively playing the company’s suite of in-app games led to the development of a new game called Shopee Farm. In this game, users log in to “water” a plant, which accumulates points and presents a prize when fully grown.

A social element is included, where users can water their friends’ plants, thus earning more rewards. Shopee Farm offers users greater rewards for long-term effort, as opposed to previous games which focused on one-time rewards.

The initiative has proven a success, with over 1 billion in-app games played in 2019. The most popular in-app games have ‘group play modes’ that offer users greater rewards when they invite and play with friends. This has been especially well-received in markets like Thailand, Indonesia, and Vietnam, where shopping is a very social affair.

These games are all localized to fit the tastes of users in each country. The same game on a Thai customer’s app might look different for people in Indonesia, for instance.

Photo credit: Shopee

Social

The final key to engagement lies in making shopping a social event. After all, users seem to already have a habit of showing off their purchases to friends and family, so why not move that experience within the app?

Enter the Shopee Feed.

The function lets users share content about their purchases and listings to the larger Shopee community. Similar to other social media platforms, Shopee Feed allows users to follow and receive real-time updates from their friends, as well as sellers and brands.

Another social feature is Shopee Live Chat. It enables direct conversations between buyers and sellers, encouraging quick discussions and clarifications before any transaction occurs. Customized stickers and media-sharing enhancements add to its social aspect.

“These conversations are a crucial part of the social experience,” notes Pan. “A whopping 1 million messages were exchanged per hour in 2019, showing how much users crave a social ecommerce experience.”

To overcome the language barrier, the firm leverages machine learning and natural language processing to develop real-time translations for the chat feature. This tech, applied in conjunction with local human expertise, helped Shopee build a translation engine that can accurately capture linguistic pairings and even local colloquialisms.

A translation model matches two or more languages together, allowing, for example, a Singaporean buyer to send a message in English, and for the Indonesian seller to receive it in Bahasa Indonesia.

“In blind BLEU tests, which are industry-standard tests for machine translations, our engine was found to score higher than leading third-party APIs,” she points out.

In a region that speaks at least nine separate languages, helping buyers and sellers understand each other is paramount. “This allows for smoother interactions when users shop for overseas products on Shopee,” says Zhou Junjie, the company’s chief commercial officer.

Zhou Junjie, chief commercial officer at Shopee / Photo credit: Shopee

Putting everything together

Beyond individual initiatives, Shopee has also introduced a livestreaming component called Shopee Live. This feature has enabled brands and sellers to better reach and engage users across the region, and it has scaled rapidly, with more than 1 million hours of content streamed in Q4 2019.

Local celebrities have been brought in to increase user adoption, engaging in a variety of live performances. These include talk shows with guests such as Taiwanese singer-actress Rainie Yang, among others.

With the internet economy expected to hit US$300 billion by 2025, ecommerce players will no doubt expand on their data and technology efforts to create an even more engaging shopping experience for customers.

“We believe that online shopping is already constantly evolving to cater to the digital lifestyles of a population that is becoming more connected and mobile,” says Zhou. “There is a need for more meaningful interactions.”

Find out how you can be a part of Shopee’s efforts to create a better shopping experience on the company’s careers page.