Can you forecast NRR, or only report it?
You can forecast it, but never as a single number. Most teams calculate NRR after the fact, put it on a board slide, and treat next year's figure as an assumption typed into a planning model. That assumption is where the plan breaks, because NRR is an output of several independent movements that each have their own driver and their own lead time.Forecast the movements. Let the ratio fall out of the arithmetic. A team that forecasts one blended retention percentage cannot answer why the number moved, and cannot act on it in time to change the outcome.
The value of any forecast is knowing the likely shape of the period early enough to do something about it. A retention number that arrives after the quarter closes tells you what happened. It does not help you.
What should an NRR forecast model separately?
Every line of the monthly ARR waterfall, with contraction split from churn. ORM builds this as a reconciling waterfall by month, where Beginning ARR always equals prior month Ending ARR. That constraint is what makes the model trustworthy, because a missing movement has nowhere to hide.| Component | Direction | Primary driver | Lead time |
|---|---|---|---|
| Beginning ARR | Base | Prior month Ending ARR | Known |
| Churned customer ARR | Contraction | Full logo loss | Contract dates plus risk signals |
| Churned product ARR | Contraction | Module or product dropped at renewal | Usage by product line |
| Product decrease ARR | Contraction | Seat or tier reduction | Headcount and utilization |
| New customer ARR | Expansion | New logos landed | New business pipeline |
| New product ARR | Expansion | Cross-sell into existing base | Expansion pipeline |
| Increase product ARR | Expansion | Seat or tier growth | Utilization against entitlement |
| Ending ARR | Result | Arithmetic | Derived |
How do you forecast each component?
Contraction from the contract base, expansion from pipeline. They are different problems and they deserve different methods.Start with the renewal base, since most of next year is already contractually visible. List every contract with a renewal date in the forecast window, then apply risk at the account level rather than a flat percentage across the base. Account-level risk uses signals you already collect, including support case volume, executive engagement, and whether anyone has touched the record. Absence is the strongest input on the retention side, exactly as it is on the deal side.
Model product decrease against utilization rather than against sentiment. An account using a fraction of the seats it pays for will right-size at renewal regardless of how much it likes you, and the finance team on the other side runs that report before the call.
Forecast expansion as pipeline with its own creation rate, not as a percentage uplift on the base. Cross-sell and seat growth need opportunities that get created, qualified, and closed, and a plan that assumes expansion appears without pipeline is a plan with a hole in it.
How do you handle cohort mix?
Weight by cohort, because a blended rate assumes a base that no longer exists. Retention differs sharply between a cohort landed two years ago under one packaging model and a cohort landed last quarter under another. When the new cohort is large relative to the base, blended history stops predicting anything.Build the forecast cohort by cohort where volume allows, then aggregate. Where a cohort is too small to be stable, group it with the nearest comparable segment and note the assumption. Segment before cohort if you have to choose. Mid-market and enterprise retention behave differently enough that mixing them produces a number describing neither.
How accurate should an NRR forecast be?
Tighter than a new business forecast, because the base is known. Forecast accuracy on new and expansion revenue typically lands near 90 percent when produced manually, at real cost in time and with no responsiveness to changing conditions. ORM targets 95 percent on that same new and expansion scope, holding from day 1 through day 90 of the quarter and updating as conditions move. Renewal sits outside that scope and has its own dynamics.Retention forecasting should beat new business on error, since renewal dates and contract values are known inputs. If it does not, the cause is usually that contraction was modeled as an average instead of account by account. Broader context on error ranges is covered in forecast accuracy.
How do you review the forecast each month?
Against the waterfall, movement by movement, with a named owner per line. Open the reconciling waterfall, compare forecast to actual on each component, and ask which assumption produced the gap. A miss driven by product decrease is a packaging and utilization problem. A miss driven by churned customer ARR is a coverage problem. The two look identical in the NRR ratio and need opposite responses.Watch the spread between gross and net retention on the same chart. When NRR holds steady while GRR falls, expansion is covering an accelerating loss rate, and the arithmetic only works until the expansion engine slows. The reporting definitions behind both figures are covered in net revenue retention, and the discipline of grading a forecast against what actually happened is covered in sales forecasting best practices.
Frequently Asked Questions
Can net revenue retention be forecast, or only reported?
It can be forecast, but never as a single ratio. NRR is an output of several independent movements, and each one has a different driver and a different lead time. Forecast the components in a monthly ARR waterfall and let the ratio fall out of the arithmetic.
What components does an NRR forecast need?
The ORM waterfall runs Beginning ARR, churned customer ARR, churned product ARR, product decrease ARR, new customer ARR, new product ARR, increase product ARR, and Ending ARR. Beginning ARR each month equals prior month Ending ARR, which forces the model to reconcile.
How far ahead can you forecast NRR?
Four quarters is workable because most of next year's renewal base is already under contract and its dates are known. The uncertain parts are expansion timing and contraction depth, so widen the range on those components rather than on the ratio itself.
Why does NRR look fine while the business is losing customers?
Expansion from a healthy segment can cover accelerating losses elsewhere. Report gross revenue retention next to NRR on the same waterfall. When the two diverge, the business is buying its retention number with upsell rather than earning it with retention.
Should renewals be forecast the same way as new business?
No. Renewals have known dates, a known base, and a bounded outcome, while new business has to be created before it can close. Model them in separate streams and consolidate at the end, or the renewal base will absorb the volatility that belongs to new business.
See how ORM turns these insights into action
ORM builds custom revenue forecast models for B2B SaaS companies. Not dashboards. Prescriptive analytics that tell you what to do next.
Schedule a Demo