is no stranger to deep tech. It has an army of sorting robots and automated systems to manage demand and supply, and some of the most sophisticated artificial intelligence
(AI)-powered recommendation engines. But the home-grown e-commerce major knows that it has to dive even deeper into technology if it wants to capture the next 100 million users, who will be mostly from the small towns.
These are the people who are only beginning to get familiar with the internet via their smartphones. So converting them into online shoppers and retaining them would depend on the kind of shopping experience the Walmart-owned company is able to offer on its platform.
Driving the effort to design these tech systems are two key persons at Flipkart: Utkarsh B, principal architect and tech advisor to CEO Kalyan Krishnamurthy, and Mayur Datar, chief data scientist. Utkarsh is also leading a new team at Flipkart
called Applied Science and Technology (AST) group. The unit is said to be working on innovations and models that will enable Flipkart’s growth in tier III towns and beyond.
Flipkart’s principal motive behind the adoption of new technology and data science is getting to know more about its customers. The more the data on users, the better will it be able to provide product recommendations, ads and offers, which in turn will make for a better customer experience.
“It is more about techniques that will allow us to go deeper in understanding target customers,” says Utkarsh. “Another interesting insight we got when we looked into the data is that different members of the same family are using one account to shop on Flipkart,” says Utkarsh. Naturally, this complicates product recommendations.
To counter this problem, his team studies product searches at particular time intervals and decodes order histories. “With the help of data we are now able to distinguish between the buying patterns in one single app and offer product suggestions accordingly,” adds Utkarsh.
The data that feeds into Flipkart
platform also helps it decide what products to show to users in specific regions. It also helps it identify the gap between what people are searching for and what’s available, which is an important input for the company’s private label strategy.
“For example, even before wearable devices were available in India, people started searching for it on Flipkart and that told us that this could be the next big trend,” explains Utkarsh. “Data helps us refine our private labels and tells us what kind of inventories to maintain.”
The data also feeds into machine learning (ML) models that determine the optimum price points for various products. The models also take into account page views, reviews and comments to decide on whether to scale up or down the inventory of these items.
Flipkart is also deploying sizeable engineering resources to further fine tune its search function. According to Mayur Datar, who leads a team of 45 data scientists, effort is underway to make the search bar on the app more effective and accurate so that it also understands queries in regional languages and colloquialisms.
“Flipkart is strong in tier II and tier III markets, and here, users sometimes search for things like ‘kala juuta’ or ‘half pants’ or ‘mehandi colour shirt’. Or, they type the wrong spelling of ‘jeans’,” explains Datar. To make sense of these queries and show up the relevant results, Flipkart is training its search algorithms and developing new language models. “The search is set for a big overhaul,” says Datar, adding that voice search too may be introduced this year.
Last year Flipkart acquired Liv.ai, a chat bot start-up, and is said to be investing to scale up its core tech.
All these new innovations are above and beyond the engineering team’s constant effort to perfect core operations like demand forecasting, warehouse management, supply chain and logistics.
In India, e-commerce is largely a fight between Amazon and Flipkart, and the player that is able to provide the better customer experience will win the game. In the case of both, technology will play a huge role in achieving that. Observers say that it is remarkable that Flipkart has managed to give a strong fight in this area to Amazon — a more mature firm with tech systems perfected in many markets around the world.
Solving India problems
Flipkart is redesigning ‘search engine’ to make robust
This will help in understanding queries in local dialects
Company preparing to launch voice search in local languages
Using AL and ML to serve better product recommendations to first-time users
Working on models to efficiently predict future demand