Abstract:- The Honest Company is one of the leading national e-commerce brands, focused on promoting healthy and happy lives. One core challenge we face is measuring ROI across the numerous marketing acquisition channels and in turn, optimizing our budget allocation. Luckily, we have large amounts of data, both structured and unstructured, which is used to learn patterns and insights about how we acquire new customers. We leverage this data to build machine learning models for smart customer segmentation, which helps the acquisition team derive maximum ROI out of every dollar spent. In this talk, we will touch on some of our machine learning approaches and how we leverage data to predict attributes like customer lifetime value and customer churn rates, which are then used to optimize spend allocation.
14. 1 4
NEW USER LTV DISTRIBUTION
Regression r_squared = 0.2 — 0.35
Besides heteroscedasticity condition was not satisfied
NEW USER LTV
15. 1 5
NEW USER LTV DISTRIBUTION
Regression r_squared = 0.2 — 0.35
Besides heteroscedasticity condition was not satisfied
NEW USER LTV
16. 1 6
New User LTV Distribution
(after removing users with low future LTV)
NEW USER LTV
17. 1 7
New User LTV Distribution
(after removing users with low future LTV)
Multi Class Problem
NEW USER LTV
18. 1 8
How to get to the dollar value?
Bootstrapping to Rescue
NEW USER LTV
19. 1 9
How to get to the dollar value?
Bootstrapping to Rescue
Multi-Class Model Probability Vector = [0.09, 0.21, 0.22, 0.20, 0.18, 0.10]
NEW USER LTV
20. 2 0
How to get to the dollar value?
Bootstrapping to Rescue
Using the class with P_max
Multi-Class Model Probability Vector = [0.09, 0.21, 0.22, 0.20, 0.18, 0.10]
NEW USER LTV
21. 2 1
How to get to the dollar value?
Bootstrapping to Rescue
Using the class with P_max
Using all classes — Wighted Bootstrapping
Multi-Class Model Probability Vector = [0.09, 0.21, 0.22, 0.20, 0.18, 0.10]
NEW USER LTV
22. 2 2
SUMMARY
• At The Honest Company, machine learning models
have improved several aspects of the business
• A model output can be used as features for final model
• Statistical inference can improve the results of a model
SUMMARY