The lifeblood of any business is repeat customers. They form a dependable revenue base to build upon over time. Without repeat customers, it becomes nearly impossible to predict future growth beyond a few quarters.
When many commerce or marketplace businesses pitch us, they flaunt their “repeat rate” — defined as the percentage of revenue (or customers) in a given period derived from users who first purchased in a prior period. Unfortunately, repeat rate is unreliable on its own. Two businesses with the same repeat rate can be drastically different in profile. Moreover, the faster you grow, the lower your repeat rate. Which would you rather have: fast growth, or higher repeats?
Both of these factors make repeat rate nothing more than a vanity metric. Let’s dig into the root cause and figure out why.
Consider two companies: Company A and Company B. They are identical in all but 2 respects. The first is what I call the “cohort/cohort growth rate” (CCGR).
This concept is a bit tricky, but imagine you have a cohort that starts in January, which purchases $100K of goods that month, and $50K of goods in February, and continues to decline by 50% per month forever. Then, imagine a second cohort of customers that starts in February, which spends $200K that month, $100K in March, and declines at the same rate.
The 100% growth of the first $200K in February over the first $100K in January is what I call the CCGR. It’s a measure of the strength of successive cohorts right out of the gate. CCGR is notably distinct from the overall month/month growth of the business, which can vary significantly even when CCGR is constant, as shown below:
The second factor that makes Company A and B different is their respective churn rates. Assume Company A has 0% CCGR and 25% churn, while Company B has 20% CCGR and 10% churn. Both start with $100K/month in revenue.
Still with me?
To analyze both companies, we’ll use a common tool: the cohort stack chart. Each color corresponds to a single cohort of customers, and the newer cohorts get stacked on top of the older ones. In the below stack chart, we can see Company A adding the same new cohort revenue each month (i.e. 0% CCGR), but flattening out due to high churn:
We can all agree that Company A is a bad business. The combination of 0% CCGR and high churn causes total revenue to flatline at $400K/month within 2 years.
Yet, Company A has a high repeat rate. The below graph shows the same revenue stack over time, simplified into new vs. repeat revenues, as well as the repeat rate, which approaches 75% in steady state:
Turning to Company B, we see a drastically different revenue ramp:
Company B is crushing it! The business is compounding revenue growth month over month, despite what most would consider high churn. While it looks markedly different than Company A, its repeat rate is the same — 75%:
The key takeaway is that repeat rate is not meaningful on its own. Instead, it is a function of both CCGR and churn.
In fact, you could run this experiment for any combination of CCGR and churn and find identical repeat rates for drastically different businesses. I’ve highlighted Company A and B in the below table, to further illustrate the point:
Furthermore, take any column and trace it south; we can easily see that CCGR is inversely proportional to the repeat rate — assuming constant churn. While a high repeat rate sounds desirable, it is more often than not an indication of sluggish growth.
In my next post, I’m going to suggest a different way of measuring repeat customer purchasing behavior, which disambiguates growth from churn. Give me a follow on Medium to find out what it is!
Authors
