पाठशाला Pathshala · हिसाब Hisāb, Unit economics · Lesson 15 · Build
Net revenue retention: the number that makes SaaS compound
Net revenue retention says whether your existing customers are a shrinking pool or a growing one. Measure it by cohort beside gross retention and learn what 100, 110 and 120 per cent do to revenue.
Pathshala, The Founder Library · 11 October 2026 · 7 min read
A software company with 120 per cent net revenue retention could stop selling to new customers tomorrow and still be twice its size in four years. One at 90 per cent must replace a tenth of its revenue every year before it grows at all. Same product category, same sales team, entirely different businesses, and the difference is one ratio most founders compute wrongly or not at all.
This lesson defines net and gross retention precisely, shows how to measure them by cohort, works a Chennai software company through both, and then shows what 100, 110 and 120 per cent mean for revenue and for what the company is worth.
Gross and net retention, defined
Take every customer who was paying twelve months ago and their annual recurring revenue at that time. That is the starting figure. Today, the same customers pay a different amount: some have left, some have moved to a smaller plan, some have added seats, modules or usage. Net revenue retention is what those customers pay today divided by what they paid then. Gross revenue retention is the same calculation with every increase left out, so it can never exceed 100 per cent; it measures only the leak.
Freshworks, the Chennai-founded company listed on Nasdaq, publishes its method every quarter, and it is the right one to copy. It takes ARR from customers as of twelve months before the period end, takes ARR from that same set of customers at the period end including upsells, cross-sells and renewals and net of contraction and attrition, and divides the second by the first. Its results for the fourth quarter of 2025 put the figure at 108 per cent. The [glossary](/library/glossary) carries the short definitions.
Measure it by cohort, not across the company
The commonest error is to divide this year’s total revenue by last year’s. That ratio mixes new customers into the numerator and calls the result retention; a company losing a fifth of its base can show 140 per cent that way. The set of customers in the numerator must be exactly the set in the denominator, and nobody acquired during the year belongs in either.
Then go one step further and compute it separately for each annual cohort: the customers who started in FY23, in FY24, in FY25. A company-wide trailing figure is a weighted average of old cohorts that have settled and young ones still sorting themselves out, and it moves when the mix moves. Cohort figures show whether the product is getting better at keeping and growing customers, which is the question. Split by plan or segment as well. Small accounts and large accounts behave differently, and a single number averaged across them describes neither.
A Chennai software company, worked
A Chennai company sells plant-maintenance software to mid-sized manufacturers. Customers paying a year ago carried ₹8 crore of ARR. In the year since, customers worth ₹80 lakh left, others downgraded by ₹40 lakh in total, and the rest added plants, users and a spares module worth ₹1.6 crore. Gross retention is ₹8 crore less ₹1.2 crore over ₹8 crore: 85 per cent. Net retention adds the ₹1.6 crore back: 105 per cent.
Both numbers matter and they say different things. The 105 is respectable; the base grows a little on its own. The 85 says that fifteen per cent of what customers paid walked out, and that expansion from the happy customers covered it. If the expansion slows, as it does when the easiest upsells are done, net retention falls toward 85 with nothing else changing. A founder who reports only the net figure is reporting the cover and not the leak.
What 100, 110 and 120 per cent mean
Net retention compounds like interest on the installed base. At 100 per cent a cohort is the same size forever and all growth must come from new sales. At 110 per cent a cohort is 1.61 times its starting size after five years and doubles in a little over seven. At 120 per cent it is 2.49 times larger after five years and doubles in 3.8. Below 100 the arithmetic runs the other way: at 90 per cent a cohort is down to 59 per cent of its starting size in five years.
Put the Chennai company’s ₹8 crore base beside ₹4 crore of new ARR sold every year and hold sales constant. In five years ARR is ₹21 crore at 90 per cent retention, ₹28 crore at 100, ₹32 crore at its current 105, ₹37 crore at 110 and ₹50 crore at 120. The sales team did the same work in every case. The difference of more than two times between the first and last is entirely what happened after the sale.
The figure opens on the Chennai company. Push expansion from 20 to 35 per cent and watch year-five ARR pass ₹45 crore. Then put expansion back and halve churn instead; the gain is smaller, but it lifts gross retention, which is the part a buyer trusts. Set new sales to zero to see what the base does alone.
Why investors pay for it
Bessemer’s Scaling to $100 Million, written from 2021 data at the height of the cloud market, put average net retention at 140 per cent for companies between $1 million and $10 million of ARR and 120 per cent from $10 million upward, with a median of 125 per cent in the smallest band. It found gross retention steady at 85 to 90 per cent across sizes. It also showed how segment shapes the number: Mindbody reached 109 per cent on contracts of about $2,000 a year, Okta 123 per cent on contracts above $50,000, and HubSpot ran at 70 to 80 per cent gross retention before its IPO while selling to small businesses. Treat these as the shape of good, not a target, and expect Indian SMB software to sit nearer Mindbody than Okta.
The reason investors pay for the number is that revenue which grows without new sales spending is worth more than revenue that must be bought again every year. David Skok’s Unlocking the Path to Negative Churn reports an investment bank’s analysis of public SaaS valuations in which growth was the top factor in the revenue multiple, with retention and upsell strong secondary factors, and a two-point rise in retention went with a 20 per cent higher multiple. The bank is unnamed, so read the figure as a direction rather than a law. The direction is consistent: a company whose base compounds at 120 per cent needs far less new selling to grow, so more of each rupee of revenue reaches cash.
New sales decide how fast a company can grow. Net revenue retention decides whether that growth is added to a pool or poured into a sieve.
Five ways NRR gets flattered
New customers in the numerator, by dividing total revenue by last year’s total. Churned customers dropped from the denominator, by measuring only customers still active. A quarter annualised, by taking one strong quarter of expansion and multiplying by four. Price rises counted as expansion without saying so; a one-off increase lifts one year’s figure and then vanishes, so report it separately. Small samples: a company with thirty customers can move ten points on one account. Skok’s SaaS Metrics 2.0 names the two honest routes to expansion, a pricing axis that grows with the customer’s use, such as seats or volume, and upsells to higher tiers or extra modules. If the expansion did not come from one of those, ask where it came from.
The quarterly retention review
In the second week of each quarter, for every annual cohort and every segment, compute four numbers from the billing system: starting ARR twelve months ago, ARR lost to customers who left, ARR lost to downgrades and ARR gained from expansion. Report gross and net retention side by side for each cohort and for the trailing twelve months, with the count of customers behind each figure.
Then read three things. Whether gross retention is rising, flat or falling cohort by cohort; that is the product and the service. Whether expansion is coming from a pricing axis that grows with use or from one-off upsells; the first compounds and the second does not. And which ten accounts moved the number most in each direction, with a name against each next step. Put the cohort table in the board pack, and in the [weekly metrics review](/library/weekly-metrics-review-one-page-one-hour) track the leading signals: seats added, usage against plan limits, renewals due in ninety days.
Nothing here is legal, tax or investment advice. The Chennai company is illustrative; benchmarks from 2021 describe a different market from today’s.
Sources
- Freshworks Inc., Freshworks Reports Fourth Quarter and Full Year 2025 Results, 10 February 2026 — Net dollar retention of 108 per cent and the method: ending ARR over entering ARR for the same customers.
- Mary D’Onofrio, Ethan Ding and Atlas editors, Scaling to $100 Million, Bessemer Venture Partners, September 2021 — Average NRR of 140 per cent at $1–10M ARR and 120 per cent above; GRR of 85–90 per cent; Mindbody, Okta and HubSpot examples.
- David Skok, Unlocking the Path to Negative Churn, forEntrepreneurs — Negative churn; an unnamed bank’s finding that a two-point rise in retention went with a 20 per cent higher multiple.
- David Skok, SaaS Metrics 2.0: A Guide to Measuring and Improving What Matters, forEntrepreneurs, January 2013 — Two routes to expansion revenue: a variable pricing axis and upsells or extra modules.