The honest answer to the benchmark question
There is no authoritative benchmark for contraction rate, so judge it against the arithmetic of your own retention rather than against a number from a survey. The test that holds up: expansion has to exceed contraction plus churn for net revenue retention to clear 100%. A company with a strong expansion motion can absorb meaningful contraction and still compound. A company with a flat product and no expansion surface cannot absorb any.That reframes the question. The useful thing to know is not whether your contraction rate is average. It is whether contraction is trending, where it concentrates, and whether it is recoverable.
Three tests worth more than a benchmark
- Concentration. Contraction spread thinly across the base usually reflects normal seat rightsizing. Contraction concentrated in one segment, one cohort, or one product is a fit problem, and it will keep going. - Direction. A stable rate is manageable at almost any level as long as expansion outruns it. A rising rate is a forecast problem regardless of the starting point. - Precedent. Track how many contracted accounts churn at the next renewal. If most of them do, contraction is not a softer outcome than churn, it is an earlier stage of it.
Separate the lines before you measure
Contraction is easy to mismeasure because it hides inside net movement. ORM separates it into three lines in its monthly ARR waterfall: Churned Customer ARR, Churned Product ARR, and Product Decrease ARR. The waterfall runs from Beginning ARR to Ending ARR, with Beginning ARR each month equal to the prior month's Ending ARR, so every movement has to reconcile.
The split matters for diagnosis. A customer dropping one product is a packaging or adoption issue. A customer cutting seats on the product they kept is a deployment issue. Both land in a blended contraction number and neither is visible there.
Predicting it early
Contraction is one of the more forecastable revenue movements because the conditions appear months ahead. Unused entitlement is the clearest input. Support engagement is another. ORM finds that customers with no support cases at all are at risk, and so are customers with 7 or more in a year, while accounts with 3 to 5 tier 2 or tier 3 tickets are less likely to churn because they are engaged and getting help.
Feed those signals into the renewal forecast and contraction becomes a number you plan for instead of a variance you explain. For the wider modeling approach, see how to forecast revenue.
Frequently Asked Questions
What is a good contraction rate for B2B SaaS?
No standards body publishes one, and the figures traded around are practitioner conventions rather than measured benchmarks. The defensible test is arithmetic: contraction is acceptable when expansion still clears it plus churn by enough to keep net revenue retention above 100%. Anything that breaks that equation is too high for your model.
Is contraction better or worse than churn?
Better in the short term and often worse as a signal. A contracted customer is still a customer, so the revenue is recoverable. But contraction usually means the account cut what it was not using, which is the same condition that precedes a churn decision at the following renewal.
How do you measure contraction rate?
Divide recurring revenue lost from downgrades, seat reductions, and dropped products by the recurring revenue at the start of the period. Exclude customers who left entirely, since that is churn. Measuring the two together produces a number that cannot be acted on, because the fixes are different.
What causes most contraction?
Seats bought and never deployed, followed by products bought in a bundle and never adopted. Both show up as unused entitlement well before the renewal, which means contraction is one of the more predictable revenue movements if usage is tracked against what the customer is paying for.
Put these metrics to work
ORM builds custom revenue forecast models that turn concepts like what is a good contraction rate? into prescriptive action for your team.
Schedule a Demo