Get Appointment

  • contact@wellinor.com
  • +(123)-456-7890

Use Cases

Data Science Use Case

Building a predictive model to identify potential buyers who are more likely to use coupons on their purchase journey.

eCommerce Predictive Model Use Case

Context

Data Science is being applied significantly in e-commerce retail, right from demand forecasting, product recommendations to inventory management and customer experience. As data technologies gain more traction, we will see more innovative applications of the AI, Machine Learning etc., in e-commerce retail.

Coupons are an integral part of an e-commerce buyer journey. Many a time, it plays an influential factor in a customer’s buying decision. Identifying buyers who are most likely to use Coupons in their purchases can help an eCommerce site in many ways. But there’s no standard solution built for it.

Our Data Science use case project leveraged Machine Learning and other data technologies and techniques to build a predictive model that can accurately predict buyers who are most likely to use coupons in their purchase journey.

The problem

The client wanted to predict the behavior of a user who during his visit (session) on the website will leave and go get a coupon from an affiliate. The client will hence assign an affiliate id to the session.

The Solution

As a solution, we built a Prediction model leveraging Machine Learning and other data technologies that predicts whether the user is going to look for the coupons or not. By this, we identify the users who come to the e-commerce website and then leave to find coupons on an affiliate website.

Conclusion

We have been able to predict with higher accuracy the behavioral patterns of the customers on the E-commerce site. Using a random forest classifier model, we are able to predict the situations in which a customer goes outside the site to look for available coupon codes. Moreover, our analysis shows that most customers look for a coupon code and when they don’t find a satisfactory coupon, they tend to exit the site to come back on a later day. This gives an opportunity for E-commerce/platforms to design effective marketing and customer engagement campaigns to boost :

a. Customer discovery
b. Customer retention
c. Conversions and Revenue.

The future scope of the models includes predicting the time spent on the website and granularly understanding the behavioral patterns. Understanding and being able to predict the time spent on the website, looking at different products, implies intent which can be leveraged by platforms to design campaigns that make buyers “pull the trigger”. Other metrics captured in the dataset include Page Depth, defined as the number of sites a customer visits on the website. Understanding the relationship between the page depth and whether the customer moves to get a coupon helps us determine the likelihood of a customer willing to buy a particular product. In essence, we can predict and give each customer a likelihood score which can then be used for targeted advertising campaigns, optimizing sales funnel, dynamic pricing based on the likelihood score, and ultimately new ways to increase the revenue of the platform.