Net revenue retention measures the growth a customer base produces on its own, and it is only as good as the cohort definition and the concentration behind it.
Net revenue retention is the recurring revenue a cohort of customers generates today divided by what the same customers generated a year ago: (starting ARR + expansion − contraction − churn) ÷ starting ARR. On an illustrative cohort starting at 50.0 million of ARR, 14.0 of expansion, 2.5 of contraction and 4.0 of churn give NRR of 115.0 per cent and gross retention of 87.0 per cent. Take out the three largest accounts and the same base retains 94.3 per cent.
Worked in full in The Growth Equity Investor by Julian R. Sterling, with every figure reproduced in a free workbook.See the book on Amazon →
NRR is the first metric a growth investor asks for in a software or subscription business, because it measures growth that needs no new sales: revenue the existing base adds on its own. A number above 100 per cent means the company would grow even if it signed no one. It is also one of the easiest numbers to flatter, through what goes into the numerator and through what the average hides.
| Movement over 12 months | Top 3 accounts | All others | Total |
|---|---|---|---|
| Starting ARR | 6.0 | 44.0 | 50.0 |
| Expansion (upsell, seats, price) | 10.0 | 4.0 | 14.0 |
| Contraction (downgrades) | 0.0 | −2.5 | −2.5 |
| Churn (lost customers) | 0.0 | −4.0 | −4.0 |
| Ending ARR of the same customers | 16.0 | 41.5 | 57.5 |
During the year the company also signs new customers worth 8.0 million of ARR. They are deliberately left out of the cohort: they were not customers at the start, so they have nothing to retain.
NRR = (starting ARR + expansion − contraction − churn) ÷ starting ARR
GRR = (starting ARR − contraction − churn) ÷ starting ARR, capped at 100 per cent by construction
In Excel, with one row per customer: =SUMIFS(ARR_End,ARR_Start,">0")/SUM(ARR_Start) for NRR, and, with a helper column Retained =MIN(ARR_End,ARR_Start) on each row, =SUM(Retained)/SUM(ARR_Start) for GRR.
NRR = (50.0 + 14.0 − 2.5 − 4.0) ÷ 50.0 = 57.5 ÷ 50.0 = 115.0 per cent. GRR = (50.0 − 2.5 − 4.0) ÷ 50.0 = 43.5 ÷ 50.0 = 87.0 per cent. The two together say the company loses 13 cents of every starting dollar and wins back 28 through expansion.
The capped formula matters for GRR computed customer by customer: a customer who expands must count as 100 per cent retained, not more, otherwise expansion leaks into a metric designed to exclude it.
The blended 115.0 per cent is true and misleading at once. The top three accounts start with 12 per cent of the ARR and produce 71 per cent of the expansion. The base grew by 7.5 million net; the top three added 10.0, so they produced 133 per cent of the net growth while everyone else shrank. Excluding them, NRR is 41.5 ÷ 44.0 = 94.3 per cent: a company whose broad base is in slow decline, carried by three customers.
The cohort definition also needs a date convention. Measure each customer's ARR at the same two points, twelve months apart, and keep customers who churned in the denominator with an ending ARR of zero. Dropping lost customers from the cohort, or measuring them at their last active month, quietly turns churn into a smaller number and pushes NRR up. With one row per customer, both errors are visible in a minute; with a summary slide, neither is.
Always ask for NRR with and without the largest accounts. A single customer's expansion can move a small company's NRR by many points. The question is not whether the 115.0 per cent is right, but how many customers it would take to lose it.
| Scenario | NRR |
|---|---|
| As reported this year | 115.0% |
| Churn doubles to 8.0 | 107.0% |
| All accounts except the top 3 | 94.3% |
| Top 3 stop expanding, the rest repeat this year | 95.9% |
The last line is the one to underwrite. It follows this year's cohort into a second year, before the new logos: if the three large accounts simply hold at 16.0 million and the rest of the base repeats its 94.3 per cent, next year's cohort ends at 16.0 + 39.14 = 55.14 against 57.5: NRR of 95.9 per cent. Nothing has gone wrong at any customer; three accounts have merely stopped buying more. Doubling churn costs 8 points; losing the expansion of three customers costs 19.
| Expansion | Losses 3.0 | Losses 6.5 | Losses 10.0 |
|---|---|---|---|
| 8.0 | 110.0% | 103.0% | 96.0% |
| 14.0 | 122.0% | 115.0% | 108.0% |
| 20.0 | 134.0% | 127.0% | 120.0% |
Compute NRR on the cohort of customers present at the start, show GRR beside it, and show both with the largest accounts removed. A high blended NRR with a sub-100 per cent tail is a concentration finding, not a retention one. The companion files for this book work a case built on exactly this pattern, a 120 per cent NRR that belongs to three customers. The same company's growth and margin feed the Rule of 40, where an NRR that falls from 115.0 to 95.9 per cent takes most of the growth term with it.
NRR includes expansion from existing customers; GRR excludes it and can never exceed 100 per cent. On an illustrative 50.0 million cohort with 14.0 of expansion, 2.5 of contraction and 4.0 of churn, NRR is 115.0 per cent and GRR is 87.0 per cent. GRR shows how much of the base the company keeps; NRR shows how much it grows.
No. NRR is measured on the customers present at the start of the period. Adding 8.0 million of new-logo ARR to an illustrative cohort that ends at 57.5 on a 50.0 start gives 131.0 per cent, which is total ARR growth of 31.0 per cent, not retention. The base contributes 15.0 points of that and new logos 16.0.
Compound it rather than multiply it. A monthly NRR of 101.0 per cent becomes 112.7 per cent over twelve months, not 112.0. Going the other way, an annual NRR of 115.0 per cent is equivalent to 1.17 per cent of net expansion a month. Always state which period a quoted NRR covers.
This article is one calculation from The Growth Equity Investor. The book takes the same case from first principles to the decision, chapter by chapter, and every figure it prints is a live formula in the free companion workbooks.
Get the book on Amazon →Free companion files
Also on Amazon UK · Amazon Germany · Amazon France · Amazon Canada
Reading guide: private equity and private markets → · All 453 articles →
If this book helped, or didn’t, a few lines on Amazon are worth more than they look: they are what the next reader goes on. Write a review. The workbook stays free either way.