analytics at adobe

I spent a summer at Adobe Technology Labs (Bangalore) with a multidisciplinary team consisting of a data scientist, Moumita Sinha, as a mentor and two co-interns. We decided to study and visualize shopping cart abandonment for the purpose of permission marketing.

keywords: collaboration, visualization, data science

We classified visitors of a prominent e-commerce site based on their likelihood to buy. We also evaluated their propensity to unsubscribe from related e-commerce emails. My unique contribution was visualizing both tendencies in conjunction, to make it possible to better target customers in real time, potentially by pushing offers based on customer characteristics. We filed two US patents titled Online Shopping Cart Analysis and Predicting Unsubscription of Potential Customers.

Please see the screen capture of a sample D3.js visualization below. The size of the bubbles indicated the number of items in the cart of a potential customer. The colors indicate our predictions. A bubble would stay on screen, moving from left to right, based on the length of the browsing session. Each bubble could be clicked to see specific customer information.

screen capture of D3 visualization

Please see my mentor describing an extension of our ideas at Adobe Summit 2016, in the section titled “Ad Wearout”. The underlying concept (in terms of data science) has evolved, however, the visualization I created is still relevant in both structure and aesthetic. In this case, the marketer is the user, and integration into her workflow and uniformity with the rest of the analytics dashboard are the factors that remain consistent and hence integral to my design.