How Amazon and Flipkart Use Data Science to Predict What You’re Going to Buy

New Delhi: Have you ever noticed online ads for a product popping up after you searched for it on an e-commerce site? Such targeted ads, designed to remind you of unfinished business, are the result of advanced data analysis, machine learning, and complex algorithms that aim to boost sales. Amazon, Flipkart, and other e-commerce players track every click you make on their sites and then predict what you are most likely to buy from them.

EY India analytics expert N. Balaji, who advises leading e-commerce companies, says identifying the target audience to whom online ads will be displayed on blogs, news sites and content streaming sites requires advanced machine learning.

Get fast cash in minutes!

The best personal loan with the lowest interest rates

Apply now

“This hyper-personalisation of ads uses consumer data to deliver ads to the right person at the right time. This is done to anticipate individual needs and improve the overall customer experience,” Livemint said.

Data analytics helps identify both loyal and new customers, using data extraction and segmentation to track browsing habits and spending patterns. This technology allows e-commerce companies to tailor their offers and promotions.

Walmart-backed Flipkart, India’s largest e-commerce player, which also owns Myntra and Jabong, analyses every click and tap during every user session to create something called a “journey” for each customer.

“User journey helps us understand how users navigate the Flipkart app and predict their next purchases,” a Flipkart spokesperson said.

For example, when a customer searches for information in product catalogs, search patterns are recorded and a customer persona is created so that when the customer returns to the website, search time is drastically reduced because the most relevant products the customer may be interested in purchasing are displayed.

When you buy a mobile phone, the system correctly matches the most likely product you’ll buy next—a mobile phone case or a screen protector—from millions of products in the catalog, an example of content-based filtering. In addition, the system can recommend products based on the persona of the person or the persona of a similar customer in the same demographic.

“While it sounds really simple, implementing this at the scale of a catalog that contains billions of products requires a phenomenal amount of software and hardware. With the advent of machine learning techniques, the recommendation engine has reached the next level of evolution,” explains EY India partner N. Balaji.

Flipkart says each of its customers has a single profile, but it creates a unique session every time a user returns to the site. All sessions are then combined to enrich the user profile and build the user’s journey on Flipkart, a company spokesperson says.