While Average Customer Purchase Latency is much better at predicting defecting customers, it can be difficult to calculate unless you use software like my app.
An alternative is to use RFM. It's a bit easier to calculate (the core is three simple sorts) and much easier to understand. RFM also gives you a lot more analysis options than Average Latency.
Using RFM for defection detection is weaker than Average Latency because RFM is more of a relative measurement (e.g. customer A is more defected than customer B) while Average Latency is more absolute (e.g. customer A has defected).
But not every system calculates Average Latency but many calculate RFM. Even old-school paper catalog marketing systems use RFM in their customer analysis. To use RFM for defection you'll want to set it up like so:
- count customers with a Recency of 1 as fully defected
- count customers with a Recency of 2 as potentially defected
- count customers with a Recency of 3 as borderline defected
Using this criteria, you'd want to use your anti-defection campaigns on customers with a Recency of 2.
(You could use a win-back campaign on Recency 1 customers but that's a different conversation)
Eric Davis
Segment your customers automatically with RFM
Segmenting your customers has always been touted as a powerful marketing tool but many stores avoid it because it can be time-consuming.
Repeat Customer Insights will automatically segment your entire customer base for you based on the valuable data Shopify has already collected for you. Ranging from 5 to 30 to over 125+ different segments using RFM and other models, you can pick how much power you want to harness.