Retention Rate
Typically calculated for one period like so:
The length of the period may be measured in months or years depending on the particulars of your product and how it is purchased and/or consumed.
Retention starts out at 100% and decays over time. Retention tends to hit a trough after a certain number of periods. See – for example – the retention curve on the right which is typical for demand deposit accounts (a.k.a. “consumer checking accounts”) in financial services. Retention goes from 100% in year 1 to 80% in year 2 to 75% in year 3 to 70% in year 4 to 60% in year 5. After 5 years, retention stabilizes and stays flat at 60%.
In a perfect world, your company would track purchasing behavior by cohort. A cohort is a group of distinct customers who entered your file (or started buying from your company) at the same point of time. Of course, most of us don’t live “in perfect.” Recently JD Powers looked at the retention rate for new buyers of cars. Here they analyzed the behavior of 117,000 new car buyers in 2003 regardless of when these buyers were last in the market for a new car. Measured in this way, the retention rate is more akin to brand loyalty. This methodology can make sense when it not possible to track customer behavior by cohort or where such tracking is highly imperfect.
The retention rate measures how effective you and your organization are at holding onto customers. Retention is a key driver of customer value and profitability. The impact of changes in retention don’t always show up in your bottom line immediately.
If you sell on a subscription model, most likely you don’t look at retention as a key metric but instead look at churn.
Best practices
Well designed retention metrics can be used for:
- Diagnostic purposes Decreases in the retention rate over time may signal a drop in product quality, a problem with your pricing, and/or a new competitor with an offering that is a better fit for the customer’s needs than your offerings.
- Predicting & Managing Customer Defections By analyzing customers who have defected you may be able to identify predictable patterns. These patterns – in turn – can be used to built predictive models that allow you to identify customers who are likely to defect early enough that you can do something about it.
Related posts:
Related posts:
