4 ms·

Yep, this is a thing. I manage a similar product. It determines if ads should appear in the funnel based on the expected likelihood and value of conversion the

• lumost 4y ago
  Do you find that this kind of optimization hurts retention? I'd expect that a decrease in retention rates would decrease long-term revenue - but that an A/B test would show a revenue increase as measuring retention will never hit statistical significance relative to revenue on a 2 week test.
  • drc500free 4y ago
    It's a good question, and definitely a dynamic we see sometimes in other parts of the marketplace - all though that's usually more driven by market dynamics than consumer behavior (e.g. publishers over-optimizing something that only works until the advertisers update their bidding strategy). In this specific case our clients are generally insurance companies, and the traffic in their funnel is not made up of current customers. The point of the predictive model is to identify people who aren't going to convert, so retention isn't relevant. Passing those consumers on to competitors who are a better fit helps the carrier recoup money, and helps the consumer find a plan they're more likely to buy. In terms of re-shopping down the road (customer lifetime could average anywhere from 3 months to 5 years from now, depending on the carrier), people generally don't visit insurance websites enough to build any strong opinion about what the UX was like - so as long as it's not spammy like a travel site, it's a fairly fresh game every time they shop for a policy.