How Machine Learning Helped Worten Achieve a 3x Increase in ROAS Using Data Driven Solutions to Boost Performance
With competition hotting up, Worten wanted to test how machine learning could help leverage their data more efficiently, drive propensity to purchase, and acquire more customers.
After a successful proof of concept (POC), the team decided to take things a step further by tackling purchase propensity in individual categories as a way to further boost efficiency. This would also allow them to create specific predictive audiences based on their campaign structure, and convert the website visitors most likely to buy.
Worten used BigQuery, machine learning, and CRMint to automate audience building in Google Ads, and create propensity models for their Sports, Computers and Home categories. The team then segmented recent website users into five types – equally distributed by volume and propensity to buy something on their next visit.