Unlike “Buy ‘Til You Die” models, Retina’s proprietary Generalized Low-Rank Model can predict customer lifetime value at or before a prospect’s first transaction. To do this, we first collect all available customer data, including estimates of RFM signals (AOV, survival rate, time between purchases). Next, our proprietary semi-supervised GLRM model denoises and imputes missing data. Finally, we simulate expected behavior at the customer-level using cleaner RFM signals.
To learn more about how our model works, check out our CLV Framework Whitepaper.