It wasn’t very long ago when a visit to a local shop to buy new clothes drew a warm smile from the shopkeeper. Chances are he knew the kind of clothes or fabric you prefer, the designs that appeal to your taste and the price range you’re comfortable with. In between sips of coffee or tea, you pick out what you want from the variety of choices he dishes out.
With the advent of e-commerce in India and rising affluence of consumers, that local shopping experience seems to have been replaced with couchsurfing through apps and website, trying to choose from the deluge of options available. Too impersonal, one could argue, except that personalisation through tech innovations is the engine that powers e-commerce today.
E-commerce users in India generate over 30 – 40 TB (Terabytes) of data daily, the equivalent of content in 50 lakh Yellow Pages books. On a typical day, e-commerce platforms receive tens of millions of visits and 100s of millions of product pageviews. E-commerce has also, increasingly, become the starting point for most product and price discovery & searches for customers. The scale is indeed mammoth.
And this mammoth data trail provides opportunity for e-commerce companies to deeply understand each & every user. From the moment a user fires up the website, or mobile site or app, thousands of lines of software code run against Machine Learning models to constantly generate user insights. And these insights are used to give customers a most personalised experience – akin to what was possible offline.
Let’s look at Search, the most common method for users to find products they want.
At Flipkart, for example, two-thirds of user searches are fairly broad, or generic, in nature such as shoes, headphones, mobiles, sarees, etc. While traditional search engines are great at answering specific questions – Nike running shoes, Moto E2 Power Black – they stumble on broad questions. Understandable, since there are different types and styles of shoes, and many types, patterns and fabrics of sarees. With so many possibilities, users often lose interest unless the right products are surfaced to them as top search results.
Which is why the need for innovation to personalise user experiences; so your Flipkart experience is different from mine, and is just what you’re looking for. And your search results will invariably be different from another user’s.
How is this possible?
The key lies in matching user insights with available products, similar to what we experienced at local shops, except that it has to be done with billions of data points and for millions of users in real-time.
With every customer activity (visit, search, product view, purchase etc) Flipkart’s machine learning algorithms help decode that customer’s shopping psyche – their demographics (age, gender, lifestage), behaviour (categories, stores, brands, price), location (pincode, locality of orders & deliveries), product attributes they like, their trust level with reviews & ratings , delivery expectations, similarity to other users etc. And these insights are then used to select the right products to show as search results when that customer triggers a new search.
If, for example, Devika, a shopper on Flipkart and a Product Manager professionally, buys Levi’s jeans for the sake of durability, but picks an unbranded belt to go with it for daily use, what would be the best products to show her when she searches for watches? Through her price affinity insight, our systems know she is looking for a watch in the value price range, her brand affinity insight for this category indicates she is interested in branded or unbranded watches, and her recent purchase history suggests that she would be interested in a watch design that matches her new attire.
The user insights used to select the right products for Devika are generated from Flipkart’s big data platform, an innovative tech platform which helps build, train and launch new ML techniques and Artificial Intelligence models to process behavioral data at scale and distill it to actionable insights.
Indian e-commerce has grown significantly over the past few years and 2016 saw the industry hit a high-water mark with 100 million registered users on Flipkart. But continuing the innovative streak to personalise experiences is a long and continuous journey, especially with our goal of bringing half a billion Indians from villages and smaller towns onto e-commerce.
Their expectations will differ markedly from customers in cities and we have to be ready. We’ve barely scratched the surface of what could be achieved using data & insights – not just for search, but other areas like discovery, product recommendations, merchandizing, and advertising. But we’ve made a good start and must keep momentum if we’re to ensure e-commerce fundamentally changes people’s lives for the good.
source: ET Tech