Retail churn is fuzzy and non-deterministic compared to subscription businesses
Corey Pearson · Corey Pearson, Custora // Data Driven NYC 23 // January 2014 (Hosted by FirstMark Capital) · Mar 3, 2014 · at 12:53
Corey Pearson, co-founder of retail predictive analytics platform Custora, discusses customer churn detection models at Data Driven NYC.
0:00 / 0:35exact quote · 35.1s
▶ Watch the full episode on YouTube → 720p mp4 · rendered on demand · StarZero watermark
“And in the world of retail, this concept of churn is really fuzzy. So Netflix knows when their customers quit, you know, because you actually cancel. But a retailer, you know, your big retail brand, Amazon or something, and you had this customer who was ordering a lot and they haven't in a couple months, maybe they're on vacation, Ah, maybe they just also started shopping elsewhere, but they're still gonna buy a little bit from you. Who knows? You know, they moved out of the country. There, there's so many things that it's very hard to have a 100% deterministic, ah, kind of mode here in our minds that says, this customer is gone.”
quote is from the automated transcript, cleaned for reading: filler sounds and stutters are removed, nothing is rephrased. names can be misheard (the analysis reads context, assessments check outside sources). how →
More from Corey Pearson
Analytics dashboards are useless unless they directly drive action
“The dashboard, the insight's never enough. You need to get either to the decision, or you need to get to the doing of the thing.”
Retailers lose 80% of customers who go inactive for 4-5 months
“What you can do is say, hey, a customer that kind of looked like this, that had this pattern, that ordered with this frequency, that bought this kind of stuff, if they go quiet for four or five months, 80% of the time, they never return.”
Algorithms cannot replace human marketers trying novel creative ideas
“There's something very, ah, helpful about having marketers in the mix there where they say, well, you know what, you know, that, that model will never realize necessarily, well, maybe we should try a funny email, or maybe we should try this type of communicati…”