In this project we take LTV as a proxy of overall financial business performance. Having done the customer segmentation, I needed to let the marketing team know how to maximize the probability of converting a customer into high LTV (eg.: giving a discount, free shipping, etc).This projects output was a list of actionables to maximize the probability of customers being high LTV.

Steps for its calculation:

Calculate each customers LTV

Took the 75% percentile as “high LTV”.

3. Having our customer database segmented into “high LTV” and “low LTV”, I ran a logistic regression to know which variables had a positive impact on high LTV, and how much the probability of being one shifted when using that specific variable (eg.: free shipping).