AI and the online shopper: the math behind personalized shopping experiences

Industry   |   
Published May 14, 2019   |   

A friend, Lisa, told me about a new clothing store online that she is absolutely obsessed with. She was awed by the simplicity of the site, and how it populated outfits that seemed to “know her taste”. That perplexed her. She was certain the site was designed for her and displayed products that attracted her attention.
Today, e-commerce has evolved to a stage where users can choose from thousands of options. However, they rarely actually buy a product from the website. The complexity of choice and the uncertainty of what’s best, leaves many online shoppers baffled. There’s just too much to choose from.
Even with search filters to narrow down choices, there are still several other factors that throw users off. Confusing product catalogs, misleading reviews and their own subjective preferences. Customers from different cultural backgrounds have different tastes. Factors like location and availability play a huge role well.
So, how exactly did this site get Lisa’s attention so easily?
Like most online shoppers, Lisa did not know that the online store’s website had‘s algorithms working behind the scenes to identify her taste and interests. And display products mapped to these interests.
Taste is the difference that taps.
The algos behind the scenes
This is how it works. is a taste-led AI platform powered by ever-learning algorithms. The algorithms first pinpointed Lias’s search location through IP tracking. They can identify a user’s city with reasonable accuracy. The algorithm also identified popular products of that city based on other users’ transaction history. This process is called collaborative filtering. The collaborative filtering algorithms predict future user actions based on previous user actions.
personalized shopping experiences
For example, if 50% of users who click on a particular t-shirt, also click on a pair of blue jeans. However, only 5% of users click on the jeans alone. So, by clicking on the t-shirt, the user indicates an affinity to the jeans, which the algorithm processes. Similar interactions across product lines help increase the accuracy of the algorithms.
Back to Lisa. notices the products she lingers on and for how long. It also takes note of her preferences like color, size, style and price. She’s then shown a personalized list of products that are reordered according to these inferences. As she continues to interact with the product list, the algorithm constantly makes note of her choices. And regularly updates the list accordingly.
personalized shopping experiences takes the shopping experience one step further. The algorithms also recommended several accessory options to the outfits Lisa picked under the widget ‘You may also like’.
And so with the help of, the online store was successful in personalizing Lisa’s end to end shopping experience to her tastes and preferences. As a result, she was hooked! And returns to the same site over and over again, to shop for her favorite clothes.