Identify high purchase propensity e-commerce visitors
A leading online jewelry chain in India wanted to identify their website visitors with a higher propensity to purchase with a view to driving conversion.
We created a Data Mart with a Customer Single View using base variables (such as age, location, pages visited, events clicked, past transaction details, queries and complaints) and derived variables (e.g. customer value index, engagement index, time spent on the website, no of media channels used).
We then created an Ensemble Model to predict visitors with high sales potential, first reducing the 600 events + 40 transactional Variables to efficient number of variable components & then created an ensemble prediction model for each segment, from logistic regression, random forest and support vector machine models.
The client segmented their website visitors & designed customized marketing activities for each segment such as higher incentives for those with a lower propensity to purchase score; subscriber retention offers, assigning concierges to reconnect with subscribers with a high propensity to churn.
The propensity model allowed the client to target their visitors more effectively. A key outcome was to identify the 30% visitors who constituted 70% of the high purchase propensity segment