Insight Series: The Secret Sauce of Live Chat Optimization

Designing a machine learning system that coaches people without stepping on their toes is, in many ways, the secret sauce of RapportBoost. We realized very early in the company’s history that we wanted to utilize a Human in the Loop system. That meant that the front line agent would be the mediator between our recommendations and what the website or app visitor ultimately saw. We needed to influence user behavior. It’s like the old adage of the tree falling in the forest - if we rendered an analysis and presented that insight to the user, and they didn’t take any action on it, did we really do the analysis in the first place? As a result, we were forced to think hard about how we would influence user behaviour.

I can’t go into all the specifics, but suffice it to say, as that user is engaging with customers, we get a window into their personality. We learn about how they think and how they tick. Likewise, we couple that with A/B Testing, different ways of conveying our messages, different placements on the screen, bold or highlighted text.

When we couple what we’ve learned about the user with A/B Testing, what we’re able to do is come up with very novel and creative ways of presenting information in order to influence their behavior. And that really is the crux of behavioral economics. It’s presenting information in a way that makes someone more inclined to act within their best interests.