CanadaRetail and e-commerce2022 – ongoingAir Miles
Machine learning for personalized email offers to loyalty collectors
Air Miles reported 4x product sales from personalized emails after moving its loyalty data and models to Databricks
Problem
The Air Miles loyalty program serves nearly 10 million active collector accounts and more than 300 retailers. Its data access was limited to five years, and building a new data pipeline took three to four months. The team wanted to know its collectors well enough to personalize offers.
Approach
The data team moved its retail consumer data onto the Databricks platform and built more than 250 machine learning models. These included product recommendation models for personalized email campaigns and reinforcement learning models to improve how collectors responded to promotional offers. The company also used machine learning for revenue forecasting, web traffic analysis and fraud detection on offer redemptions.
Results
Takeaway
Air Miles tied its marketing gains to a data base that held 30 years of history instead of five. The company did not say how the sales and response figures were measured.
Categories
- Industry
- Retail and e-commerce
- Business function
- Marketing and sales
- Innovation type
- Product and service innovation
- Adoption
- Off-the-shelf tool or platform
- Evidence
- Vendor case study
Sources
Source summaries
Databricks describes how the Air Miles program moved its collector data to the Databricks platform and built more than 250 machine learning models. It reports a 4x increase in product sales from personalized email campaigns and a 2x increase in collector response rates. It also gives volumes such as 100 million personalized offers a month and pipeline build times cut from 3 to 4 months to 2 days.
Full details are in the original sources.
