In +10 years running SaaS companies I have sat through hundreds of board meetings, and the ones that went badly usually had the same root cause: 6 people in a room using the same 3-letter acronym to mean 4 different things. Annual recurring revenue (ARR), churn, net revenue retention (NRR), customer acquisition cost (CAC), lifetime value (LTV) and payback are not hard. They are just rarely defined out loud.
This is the glossary I wish someone had handed me at my first board. Each metric in 1 sentence, the formula in words, the 2026 benchmark ranges from the 3 datasets I trust most, and the 5 mistakes I see boards make with them. The companies I run add up to €10M+ in ARR, so these are the numbers I look at every month.
Key takeaways
- ARR is the annualised value of your active subscriptions, and it is a run-rate for planning, not revenue you have earned, so never put it in the same column as accounting revenue.
- According to the 2026 Aleph and Benchmarkit report on 342 B2B SaaS companies, median net revenue retention is 102% and median gross revenue retention is 84%, which means the typical company barely grows from its existing customers.
- The same report puts median CAC payback at 16 months, with the top quartile under 6 months and the bottom quartile above 24, so payback tells you more about efficiency than any LTV to CAC ratio.
- The 5 board mistakes are mixing bookings with ARR, quoting NRR without GRR, using logo churn for revenue decisions, trusting LTV built on 2 years of data, and comparing against the wrong benchmark segment.
What is ARR and why do boards start there?
Annual recurring revenue (ARR) is the yearly value of all subscription contracts active today, and boards start there because it is the cleanest single measure of the size of a subscription business.
The formula in words: take every active subscription, convert it to its yearly price, add them up, and exclude anything that does not repeat, such as onboarding fees, one-off services or usage that is not contracted. Monthly recurring revenue (MRR) is the same number divided by 12, and it is the better unit for companies selling monthly plans to small customers.
The important discipline is what ARR is not. It is not revenue in the accounts, which is recognised as the service is delivered. It is not bookings, which are contracts signed but possibly not yet live. And it is not cash, which arrives when customers pay, sometimes a year in advance. In my companies every board deck shows ARR, recognised revenue and cash collected as 3 separate lines, because they can move in different directions for a quarter.
What is churn and which churn should I track?
Churn is the share of customers or recurring revenue you lose in a period, and the churn a CEO should track for decisions is revenue churn, because losing 10 small accounts and losing 1 large one are very different events with the same logo count.
There are 3 versions. Logo churn: customers lost in the period divided by customers at the start. Gross revenue churn: recurring revenue lost from cancellations and downgrades divided by recurring revenue at the start, with no credit for upgrades. Gross revenue retention (GRR) is simply 100% minus gross revenue churn, capped at 100% by definition.
GRR is the number I trust most because it cannot be flattered. According to the 2026 Aleph and Benchmarkit benchmarks, median GRR across 342 B2B SaaS companies fell to 84%, meaning the typical company loses 16% of its recurring revenue base every year before selling anything new. SaaS Capital's 2026 survey of bootstrapped companies between $3M and $20M in ARR shows a healthier 91% median. Below 80%, growth is filling a bucket with a hole in it.
What is net revenue retention and what is a good NRR in 2026?
Net revenue retention (NRR) is the recurring revenue you keep from last year's customers after churn, downgrades and expansion, and in 2026 a good NRR for B2B SaaS is above 110%, with the median at 102%.
The formula in words: take the recurring revenue from the customers you had 12 months ago, subtract what they cancelled or downgraded, add what they expanded, and divide by what they paid 12 months ago. Above 100% means your existing base grows on its own; below 100% means you must sell just to stand still.
According to the Aleph and Benchmarkit 2026 report, based on full-year 2025 data, median NRR is 102%, the top quartile is at 110%, the bottom quartile at 92%, and 120%+ is best in class. Pricing model matters a lot: usage-based companies post a median of 108% against 98% for seat-based, and companies under $5M in ARR sit at 94%. SaaS Capital reports a 103% median for bootstrapped companies. ChartMogul's December 2025 retention report, built from 3,500 mostly smaller companies, shows a much lower 82% median for B2B SaaS and 48% for AI-native products: benchmarks depend on who is in the sample.
In my companies NRR is the metric I watch most closely, because it compounds. A business at 110% NRR doubles its base from existing customers in about 7 years without a single new logo. A business at 90% must replace 10% of itself every year before it grows.
What are CAC, LTV and CAC payback?
Customer acquisition cost (CAC) is what you spend in sales and marketing to win 1 customer, lifetime value (LTV) is the gross profit that customer generates before leaving, and CAC payback is the number of months of gross profit it takes to recover the CAC.
CAC in words: total sales and marketing cost in a period, salaries included, divided by the number of new customers won in that period. LTV: average recurring revenue per customer per year, multiplied by gross margin, divided by annual revenue churn. CAC payback: sales and marketing cost of the prior period, divided by the new ARR added multiplied by gross margin, times 12. That last formula is the one the Aleph and Benchmarkit report uses; copy it exactly so your board can compare.
The benchmarks: according to the 2026 report, median CAC payback is 16 months, improved from 18 in 2024, with the top quartile at 6 months or less and the bottom quartile at 24 or more. Deal size drives it: companies selling under $5,000 a year recover CAC in a median of 11 months, while those selling $50,000 to $100,000 contracts take 22. The classic rule of an LTV to CAC ratio above 3 still circulates, but I use it less every year: LTV rests on a churn assumption projected 5 to 10 years out, payback only on numbers you already have.
Which 2026 benchmarks should a CEO actually use?
Use the benchmark that matches your stage, deal size and funding model, because the median for a venture-backed enterprise vendor and the median for a bootstrapped company selling €50 plans differ by 20 points or more on the same metric.
These are the 2026 reference points I keep on 1 page, with their sources, so that nobody in my boards quotes a number without knowing where it came from.
- NRR: median 102%, top quartile 110%, best in class 120%+ (Aleph and Benchmarkit, 342 B2B SaaS companies, FY2025 data).
- NRR by pricing model: 108% usage-based versus 98% seat-based; 94% for companies under $5M ARR (same report).
- GRR: median 84% (Aleph and Benchmarkit); 91% for bootstrapped $3M to $20M ARR companies (SaaS Capital 2026).
- CAC payback: median 16 months, top quartile 6 or less, bottom quartile 24 or more; 11 months under $5K deals, 22 months for $50K to $100K deals (Aleph and Benchmarkit).
- Growth: median 15% for bootstrapped companies in 2026, down from 20% the year before, with the 90th percentile at 42.3% (SaaS Capital).
- Small and AI-native products: B2B SaaS median NRR 82%, AI-native 48%, with AI plans above $250 a month retaining far better than plans under $50 (ChartMogul, December 2025).
What are the 5 mistakes boards make with SaaS metrics?
The 5 mistakes are counting bookings as ARR, celebrating NRR while hiding GRR, deciding on logo churn instead of revenue churn, trusting an LTV built on a 2-year-old company's churn, and comparing against a benchmark from a different segment.
I have seen each of these turn a good quarter into a bad decision. They are definition problems, not fraud problems, and the fix is always the same: write the definition on the slide.
- Bookings as ARR: a signed contract that goes live in 4 months inflates ARR today and disappoints in the quarter it should have shown up.
- NRR without GRR: 115% NRR with 75% GRR means a few large accounts expanding hide a base that is leaving; always show both.
- Logo churn for revenue decisions: 2% monthly logo churn concentrated in your smallest plans is a pricing question, not a product crisis.
- LTV from young cohorts: a company with 24 months of history cannot know its 5-year churn, so LTV becomes a wish; use CAC payback and GRR instead.
- Wrong benchmark segment: comparing a €30-a-month product with 20+ countries of self-serve customers against enterprise NRR medians guarantees the wrong conclusion.
You do not need 40 SaaS metrics. You need 6, defined in writing, tracked against the right benchmark segment and shown together so they cannot hide each other: ARR next to recognised revenue and cash, GRR next to NRR, CAC payback next to growth. That is the whole system I use across the companies I run, and the one I ask of every founder building towards €1M ARR inside my Aurum VOS project. Write the definitions down, pick your benchmark, and let the numbers argue in the open.
Frequently asked questions
- What is a good net revenue retention rate for SaaS in 2026?
- According to the 2026 Aleph and Benchmarkit report on 342 B2B SaaS companies, median NRR is 102%, top quartile is 110% and 120% or more is best in class. Usage-based companies post a higher median of 108% against 98% for seat-based pricing, so compare against your own pricing model.
- What is a healthy CAC payback period?
- The 2026 median is 16 months, with the top quartile recovering CAC in 6 months or less and the bottom quartile taking 24 or more. It depends on deal size: under 12 months is normal for self-serve plans under $5,000 a year, while enterprise deals of $50,000 to $100,000 take about 22 months.
- Should I use LTV to CAC or CAC payback?
- Use CAC payback as the primary efficiency metric and treat LTV to CAC as a sanity check. Payback uses only numbers you already have, sales and marketing spend, new ARR and gross margin, while LTV depends on a churn assumption projected years into the future that young companies cannot know.