Using your Customer Purchase Latency is a good way to detect defecting customers but it has its drawbacks.
Especially with products that are very durable and have long buying cycles.
An example I've used in the past are my running sandals. Unlike regular running shoes, they have lasted for years so the purchase latency will be high.
Using my own purchase history, let's figure out the latencies between purchases:
- 2013ish, let's say July 25th to make it easy.
- July 25, 2014
- August 12, 2015
- December 15, 2015 (accessories)
- August 06, 2016
- April 14, 2019
- 1st to 2nd: 365 days
- 2nd to 3rd: 383 days
- 3rd to 4th: 125 days (accessories)
- 4th to 5th: 235 days
- 5th to 6th: 981 days
This latency data shows pretty clear behavior. Even more so if you ignore the accessory purchase and only look at the full sandal purchases (making the 4th to 5th latency 360 days).
With 360-385 days between purchases, that last one looks really odd. Almost as if I defected for two years during 2017 and 2018.
But in reality, since the products are so durable I was still using the prior three pairs and didn't need to order another until 2019 (my buying behavior shifted once I had enough product on hand).
With metrics like this, what could this store do with such a long rebuying cycle?
Before they change anything they should look at how the first few purchases look for the rest of their customers. My buying behavior might be an outlier and the regular average behavior is different.
But let's assume other buyers have the same annual buying behavior.
Then the store can use that to adapt their marketing campaigns.
For example, focus heavily on education or entertainment until close the year mark and then shift into messages about replacing worn sandals or talking about features on newer models. Keep hard-sells out of the campaigns and focus on durability and stability instead.
A story (or two) from a customer and how much they've used the product and how long it's latest would be a strong message to reinforce.
They could even collect these stories by asking customers for details on their usage. This would be more than a product review and could find out other things like running preferences, distances, other brands they use, etc.
That could then be combined with the behavioral data to boost their marketing more or even explore new product lines.
This isn't limited to repeat customer marketing either. e.g. the sales messages to new customers could be: "These will last for years, many customers have put 700, 1000, or even 2,000 on them before needing to replace them"
That's the magic of analyzing this sort of data. It takes a bit to slice and dice it all, but the impacts can be far-ranging.
Calculating the latency for just one customer's orders is pretty easy. Manual, but easy.
Doing the same for your entire customer base and aggregating that data into something meaningful? That's tough and needs a computer.
That's why I built the Latency Report in Repeat Customer Insights. It's such a powerful report but troublesome to create by hand (not to mention a huge time-sink).