Cohorts & Revenue Retention
Track groups of customers over time, by head count and by dollars.
In your head, jot if needed · no calculator why?
Opens at level 39. You're level 1. You can read and practice here now.
the idea
A cohort is a group of customers who started in the same period, followed over time. Always divide by what that cohort started with.
Revenue retention follows the cohort's monthly revenue instead of its head count. Net revenue retention adds upgrades and subtracts downgrades and cancellations, so it can top 100%. Gross revenue retention leaves the upgrades out, so it never does. New customers count in neither.
Heads and dollars can move apart: if the customers who stay pay more, revenue can grow while the count falls.
techniques
Ending revenue over starting revenue
- Start from the cohort's starting monthly revenue.
- Net: add upgrades, subtract downgrades and cancellations. Gross: subtract only.
- Divide by the starting revenue and write it as a percent.
worked example
A cohort starts the year paying $40,000 a month. Upgrades add $8,000 of monthly revenue, downgrades remove $2,000, and cancellations remove $4,000. New customers are not counted. What is net revenue retention, as a percent?
- $40,000 + $8,000 − $2,000 − $4,000 = $42,000.
- $42,000 ÷ $40,000 = 1.05 = 105%.
- Gross leaves out upgrades: $34,000 ÷ $40,000 = 85%.
Answer: 105%
Weight each month by its chance
- Month 1 counts in full.
- Each later month multiplies in the chance of staying. At 50%: month 2 is 50%, month 3 is 25%.
- Multiply each month's contribution by its chance, then add.
worked example
A customer contributes $200 a month while active. They are active in month 1. Assume they stay each later month with probability 50%; once gone, they don't return. What is the expected contribution over the first 3 months?
- Chances active: 100%, 50%, 25%.
- $200 + $100 + $50 = $350.
Answer: $350
watch out for
- Leaving the upgrades out of net revenue retention. That gives gross revenue retention instead.
- Using the same chance for every later month. Staying two more months at 50% each is a 25% chance, not 50%.
- Dividing one cohort's active count by both cohorts' starting sizes added together.
- Expecting customer retention and revenue retention to match. One counts heads; the other weighs each customer by what they pay.
practice
Net revenue retention
worked example
A group of customers starts the year paying $550,000 a month. Over the year, upgrades add $212,500 of monthly revenue, downgrades remove $63,000, and cancellations remove $83,500. New customers are not counted. What is net revenue retention, as a percent?
Answer: 112%
- ($550,000 + $212,500 − $63,000 − $83,500) ÷ $550,000 = $616,000 ÷ $550,000 = 112%.
Gross revenue retention
worked example
A group of customers starts the year paying $1,000,000 a month. Over the year, upgrades add $179,500 of monthly revenue, downgrades remove $73,500, and cancellations remove $30,500. New customers are not counted. What is gross revenue retention (upgrades not counted), as a percent?
Answer: 89.6%
- ($1,000,000 − $73,500 − $30,500) ÷ $1,000,000 = $896,000 ÷ $1,000,000 = 89.6%.
- It can never be above 100%.
Expected contribution over a few months
worked example
A new customer contributes $780 in each month they are active. They are active in month 1. Assume that each month after that they stay with probability 80%, and once gone they don't come back. What is the expected total contribution over the first 2 months?
Answer: $1,404.00
- Chance still active by month: 100%, 80%.
- Expected: $780 + $624 = $1,404.
- This is contribution, not revenue, and it isn't discounted.
Read a cohort
worked example
The August cohort started with 60 customers, and 51 were active in month 2. The September cohort started with 40, and 15 were active in month 4. What share of the August cohort was still active in month 2, as a percent?
Answer: 85%
- Use the cohort's own starting size: 51 ÷ 60 = 85%.
Heads versus dollars
worked example
Total monthly revenue is up 20% on a year ago, yet 26% of the customers you started the year with have cancelled, and those who stayed pay the same as before. What best explains this?
- Prices must have been cut
- New customers hid the losses from the starting group
- The two numbers must be a reporting error
- Churn was calculated using the ending customer count
Answer: New customers hid the losses from the starting group
- Customer retention counts heads; revenue retention weighs each customer by what they pay, including upgrades and downgrades.
- The starting group shrank by 26% with no upgrades, so all of the 20% growth came from new customers.