We will follow one year of a made-up business, Narmada Analytics, from the first day to the last, and compute two numbers at the end. Same data, two different stories.
Day one
On 1 April, Narmada has 100 customers and annual recurring revenue (ARR) of ₹80 crore. This is the opening base. Retention measures only what happens to this group over the year. Customers who sign later do not count, no matter how big.
What happened during the year
Over the next twelve months, four things occur to customers who were already on the books.
- Eight customers leave entirely, taking ₹6 crore of ARR with them. This is churn.
- Some others stay but cut seats or move to a cheaper plan, reducing ARR by ₹2 crore. This is contraction, sometimes called downgrade.
- A group of customers buys more users and a second module, adding ₹14 crore. This is expansion.
- Separately, the sales team signs 20 new customers worth ₹12 crore. These are new logos and sit outside retention.
Calculating gross retention
Gross revenue retention (GRR) asks: of the money we started with, how much is still here, ignoring any growth from the same customers? Start with ₹80 crore, subtract churn of 6 and contraction of 2. Result: ₹72 crore. Divide by 80: 90 per cent.
GRR can never go above 100 per cent, because expansion is deliberately left out. It is a measure of leakage. A GRR of 90 per cent says that a tenth of the opening revenue disappeared during the year.
Calculating net retention
Net revenue retention (NRR) takes the same ₹72 crore and adds expansion of 14. That gives ₹86 crore, and 86 divided by 80 is 107.5 per cent. The same group of customers, as a whole, pays 7.5 per cent more than at the start, even though some left.
The new ₹12 crore from fresh logos plays no part in either figure. Total ending ARR is 86 + 12 = ₹98 crore, but that is a different number from NRR and people confuse them constantly.
Same year, second reading
Now change one thing. Suppose churn had been ₹10 crore instead of ₹6, but expansion had been ₹18 crore. Net retention is (80 − 10 − 2 + 18) ÷ 80 = 86 ÷ 80, so still 107.5 per cent. Gross retention falls to 68 ÷ 80 = 85 per cent. NRR is identical; the health of the base is not. In the second version, one customer in every few was leaving or shrinking and a handful of big expanders covered for them.
This is the whole reason to track both. NRR tells you whether the base grows by itself. GRR tells you how sticky the product is. A company can post a lovely NRR while its GRR deteriorates, until the expanders run out of room to expand and the hole becomes visible.
Things to watch for in how it is reported
Definitions are not universal. Check the window (trailing twelve months, or a single quarter annualised), whether contraction is included in GRR, whether price increases count as expansion, and whether the measure is by revenue or by customer count. Logo retention, the share of customers rather than rupees kept, is a separate figure and often looks quite different. A company with a few giant customers can have excellent logo churn and terrible revenue retention, or the reverse.
Also look at segment splits. Enterprise customers often retain better than small ones. A blended 90 per cent could be 96 per cent in large accounts and 80 per cent among small businesses. The blended number hides an important decision about where to invest.
Finally, remember who is asking. Investors and boards often read NRR above 100 per cent as a sign of a durable business, and many consider the high-90s to be strong for enterprise software, though benchmarks differ by customer size and are only rough guides. Do not chase a benchmark without checking it against your own segments.
Figures and names in the examples are illustrative.
Leave a Reply