A major online retailer with a flash-sales business model needed better analytics than they currently had. Specifically, they needed predictive information about their customers that could be used to suggest products to buy—instantaneously and engagingly with each online visit. The purpose, of course, was not only to make sales, but to accelerate them.
Our solution consisted of aggregating our client’s data, learning from it, and developing a recommendation engine. We started by analyzing their system’s interaction with each customer and customers in general, and classifying what we learned into customer categories and buying preferences—information that lends itself to predictions. A customer shopping for shoes, for instance, would be presented with footwear in line with their preferences as inferred by the system (based on categories we derived using a classification algorithm). So, the engine might offer one buyer a selection of casual loafers, while another would see a selection of business footwear. It could also predict if the buyer was going to leave/buy within a specific period of time (one minute, five minutes, one day, and so on).
The analytical capability we introduced into our client’s system not only increased sales, up-selling, and cross-selling, but also gave their marketing staff a means to implement further conversion strategies across their website.