Case Study: Epsilon increases customers' marketing ROI with H2O.ai
Key results
The challenge
Epsilon's Abacus unit builds custom modeled consumer lists for each client marketing campaign, amounting to more than 100,000 models a year across roughly 8,000 campaigns. Its long-standing SAS-based automated system faced scale and throughput challenges processing terabyte-sized data sets. Model interpretability was also critical to prove list selections were unbiased amid increasing privacy regulation.
The solution
Epsilon partnered with H2O.ai, selected for its ability to directly support Epsilon's algorithms, process large data sets, and explain models. Jobs are queued and multiple H2O.ai models are built for each campaign, mirroring the prior SAS workflow while adding transparent, interpretable list selection.
“In the environment of increasing privacy regulations, the tools Epsilon employs need to provide transparent list selection. H2O.ai help us meet this challenging and critical need.”
ATAndrea ThorntonVP of Analytics, Epsilon
The results, in context
Moving from SAS to H2O.ai improved list response rates by 3-5%. A 3% gain equates to roughly 15,000 more relevant customers in every marketing campaign. For one large gifts brand, a 1.10% increase in direct-mail response rate drove an incremental $9.0M from a single campaign.