A multi-channel beauty brand struggled with “one and done” purchase behavior despite having a successful customer loyalty strategy after a second purchase occurred. The brand had significant intelligence on customer purchase behavior through transactional data sets as well as rich site-side data from their e-commerce portal but wasn’t harnessing any of this intelligence to influence outbound marketing through paid or unpaid channels to drive repeat purchase among their customer base.
Actable built a bespoke Machine Learning Model, which synthesized data from various sources into learning signals, and then scored individual customers based on their propensity to make a second purchase. Example variables included:
- Products purchased on first purchase
- On-site behavior patterns
- Media exposure / referrals
- Sales channels / locations
The scoring model was then deployed to a virtual machine in the client’s cloud environment, enabling programmatic daily updates.
Scores were syndicated programmatically to downstream marketing platforms, including their ESP and social media platforms, and updated daily.
Targeted messages were then sent to 1st purchasers via email and social channels in an a/b split to measure the messaging and investment strategy against high-scoring vs. low-scoring customers
- Customers who received “high” scores (top 30%) converted nearly 2X the rate of the “low” score group.
- Media investments now targeted to high-scoring customers for cost-savings and efficiency gains.
- Model being deployed across high-cost channels (e.g. direct mail) to drive further efficiency.