Net revenue retention explained for founders
What NRR actually measures, how to calculate it from a cohort, and why investors treat one number as a proxy for product-market fit.
THE SHORT ANSWER
Net revenue retention (NRR) is the revenue you keep from existing customers: take a cohort's revenue today and divide it by that same cohort's revenue a year ago. Above 100% means expansion revenue outruns churn and downgrades, so the business grows even with zero new sales. Investors read NRR as product-market fit expressed in one number.
What is net revenue retention, exactly?
Net revenue retention answers one question: if you stopped selling to new customers today, what would happen to your revenue over the next 12 months?
Take every customer you had exactly one year ago. Add up what they paid you then. Add up what those same customers pay you now, including upgrades and minus downgrades and cancellations. Divide the second number by the first. That is NRR.
New customers signed in the last 12 months are excluded. NRR measures only what happens to revenue you already had, which is why it isolates retention and expansion from sales performance.
How do you calculate NRR? A worked example
Start with a cohort of customers worth $100,000 in ARR twelve months ago. Over the year:
- Expansions (seat growth, plan upgrades) added $18,000
- Downgrades removed $5,000
- Churned customers removed $8,000
The cohort is now worth $100,000 + $18,000 - $5,000 - $8,000 = $105,000.
NRR = $105,000 ÷ $100,000 = 105%.
The formula in one line:
NRR = (starting cohort ARR + expansion - downgrades - churned ARR) ÷ starting cohort ARR
Note what happened in the example: this company lost 13% of the cohort's revenue to churn and downgrades, yet still posted an NRR above 100% because expansion covered the hole. That is both the power and the trap of the metric, and we will come back to the trap.
NRR vs GRR: when does each matter?
Gross revenue retention (GRR) is the same calculation with expansion removed. It only counts what you kept, never what you grew.
| Metric | Formula | What it tells you | Ceiling | |---|---|---|---| | GRR | (start - downgrades - churn) ÷ start | How leaky the bucket is | 100% | | NRR | (start + expansion - downgrades - churn) ÷ start | Whether existing accounts grow on net | None |
In the worked example above, GRR is $87,000 ÷ $100,000 = 87%, while NRR is 105%. Same company, two very different stories.
Use GRR when you want the honest picture of retention quality: it cannot be rescued by a few big upsells. Use NRR when you want the growth picture: it tells you whether the installed base is an engine or an anchor. Investors doing diligence will ask for both, precisely because NRR alone can hide a leak.
What NRR do investors want to see?
Plain-terms benchmarks for B2B SaaS:
- Below 100%: the bucket leaks. Every new sale partly refills lost revenue, so growth gets more expensive every quarter.
- 100 to 110%: healthy. Existing customers hold their value and grow slightly.
- 110%+: what most investors want to see before they get excited. The base compounds on its own.
- 120%+: the territory that commands premium valuations, because revenue grows meaningfully with no sales effort at all.
110%the NRR bar most SaaS investors look for
Why the obsession? Because NRR compounds. A company at 120% NRR doubles revenue from its existing base alone in under four years, before counting a single new logo. A company at 90% has to replace 10% of its revenue every year just to stand still.
Context matters when you compare yourself to these numbers. Enterprise-focused companies with seat-based pricing tend to run higher NRR than SMB-focused ones, because larger accounts have more room to expand and churn less often. If you sell to small businesses, an NRR a few points lower than an enterprise peer is not automatically a worse business.
How can logo churn hide inside good NRR?
NRR is a revenue-weighted average, and averages hide distributions.
Imagine 50 customers worth $100k total. Your two biggest accounts expand aggressively and add $20k. Meanwhile ten small customers, worth $12k combined, cancel. NRR: ($100k + $20k - $12k) ÷ $100k = 108%. Looks great.
But you just lost 20% of your logos. If either whale sneezes next year, the number collapses, and the ten cancellations may be telling you something the expansions are drowning out: the product is failing a whole segment.
The reverse also happens. A company can post 97% NRR while keeping 95% of its logos, simply because it has no expansion motion yet. That business has strong retention and an unpulled pricing lever, which is a much better position than the 108% company above.
Do this this week: calculate your NRR and your logo retention side by side for the same 12-month cohort. If NRR is above 100% but logo retention is below 90%, read your churned accounts' cancellation reasons before you celebrate the NRR number.
What are the three levers that move NRR?
Every point of NRR comes from exactly three places. Work them in this order.
Lever 1: reduce churn
Churn is subtracted straight off the top of the formula, so every dollar saved is a full dollar of NRR. It is also the cheapest lever: catching an at-risk account 60 days before renewal costs a phone call, while replacing that account costs a full sales cycle. Start with the accounts showing declining usage; behavior degrades before the cancellation email arrives.
Lever 2: expand accounts
Expansion is the additive term. Seat growth, usage tiers, and cross-sell of a second product all count. The practical move for a 50 to 500 account company: identify the customers using you most heavily relative to what they pay, and put a proactive upgrade conversation on the calendar. Expansion revenue that arrives without a discount is the highest-margin revenue you will ever book.
Lever 3: price on value
If customers get more value from your product every year (more data, more usage, more seats) but pay a flat fee, your pricing model donates your NRR back to them. Align at least one pricing axis with the dimension of value that naturally grows. This is why usage-based and seat-based companies structurally outrun flat-fee ones on NRR.
What should you show the board monthly?
Keep the retention slide to four numbers and one sentence of narrative:
| Line item | Why the board needs it | |---|---| | NRR (trailing 12 months) | The headline compounding number | | GRR (trailing 12 months) | Proof the NRR is not masking a leak | | Logo retention | Catches the whale-dependency failure mode | | Expansion vs churned ARR, in dollars | Shows which lever moved the number |
Report all four on the same cohort definition every month, and resist the urge to switch to a friendlier definition in a bad quarter. Boards forgive a soft number; they do not forgive a moving one.
If you can only track one thing between board meetings, track the leading indicator instead of the metric itself: which accounts are showing usage decline right now. NRR is a lagging scoreboard. By the time it drops, the churn that caused it happened two quarters ago.
How often should you measure NRR?
Measure monthly, on a trailing 12-month window. A shorter window (annualizing one month's retention) makes the number jump with every large renewal or cancellation, and a jumpy metric teaches your team to ignore it.
Two rules keep the measurement honest:
- Fix the cohort at the start of the window. The customers counted are exactly those active 12 months ago, no additions and no substitutions.
- Count revenue changes when they take effect, not when they are signed. A downgrade agreed in March that starts in June belongs to June.
Early-stage companies with fewer than roughly 30 customers should treat NRR as directional only. In a cohort that small, one account can swing the number by ten points, so read the individual account movements instead of the ratio.
Related resources
RELATED
01 ·Churn rate formula: customer, revenue, and cohort methods
The basic churn rate formula, plus the revenue and cohort variants that catch what logo churn hides. Worked examples you can copy into a spreadsheet.