Renewal forecasting should be the easy half of the revenue forecast. You know the contract dates, the amounts, and the customers a year in advance. Most teams still get it wrong, because they forecast renewals with a CRM stage field and a customer success manager's gut feel rather than with the usage and support signals that actually predict the outcome.
The fix has three parts: separate the renewal forecast from the new business forecast, build it on a reconciling ARR waterfall, and feed it signals that arrive before the renewal conversation starts.
Why should renewals be forecast separately?
Because renewal revenue and new business revenue are driven by different things and break in different ways. New and expansion business responds to pipeline creation, win rate, deal size, and cycle length. Renewal revenue responds to product usage, support experience, and whether the person who bought the product is still employed there.Blending them produces a total that cannot be diagnosed. A quarter where new business ran hot and renewals ran cold looks identical on the roll-up to a quarter where both were average, and the two situations demand completely different responses.
Keep the accuracy targets separate too. The forecast accuracy figures ORM publishes, around 90 percent as a typical manual result and 95 percent as the target without manual adjustment, apply to new and expansion business specifically. Renewals are a distinct forecasting problem and deserve their own tracked accuracy history.
What structure should the renewal forecast use?
A monthly ARR waterfall that reconciles, where each month's beginning ARR equals the prior month's ending ARR. The reconciliation constraint is what makes the model honest. Any movement you cannot classify shows up as a gap you have to explain.| Line | Direction | What it captures |
|---|---|---|
| Beginning ARR | Opening | Prior month ending ARR |
| Churned customer ARR | Contraction | Full customer loss |
| Churned product ARR | Contraction | A product removed, customer retained |
| Product decrease ARR | Contraction | Seat or tier reduction |
| New customer ARR | Expansion | Net new logos |
| New product ARR | Expansion | Cross-sell into existing customers |
| Increased product ARR | Expansion | Upsell within an existing product |
| Ending ARR | Closing | Sum of the above |
Which signals predict renewal outcomes earliest?
Support case volume, read at both extremes rather than as a linear scale. This is the most counterintuitive finding in retention data and one of the most useful.A customer with no support cases at all is at risk. Silence reads like satisfaction, and it usually means nobody is in the product. A customer with seven or more cases in the last year is also at risk, because that volume signals real friction. The healthy band sits in the middle. Three to five cases, typically tier two or three rather than severe, correlates with an engaged customer who is using the product, getting help, and staying.
Plotting renewal outcomes against annual case count on your own base takes an afternoon and usually reproduces the pattern. Once you have it, the ticket count becomes a leading input to the renewal forecast rather than a support metric nobody in revenue looks at.
What else belongs in the risk model?
Signal absence, sponsor change, and product adoption breadth.Signal absence works the same way in renewals as it does in new business. A committed deal with no buyer contact is a bad sign, and a renewal ninety days out with no scheduled conversation, no usage growth, and no support interaction is the same bad sign wearing different clothes. Build the report so silence surfaces automatically rather than waiting for someone to notice.
Sponsor change is the fastest-moving risk. When the person who bought the product leaves, the renewal reverts to a new sale with none of the qualification work done. Track it as an event, not as a field someone updates quarterly.
Adoption breadth matters more than adoption depth for multi-product accounts. A customer using one product heavily is easier to replace than a customer using three products moderately, and the waterfall lines for churned product ARR and product decrease ARR will show you which accounts are drifting toward the first case.
How do you forecast the renewal number month by month?
Segment the renewal cohort by risk band, apply historical retention rates by band, and forecast contraction lines separately from full churn.Start with the known contract base for each month. Assign each account to a risk band using the signals above. Then apply the historical outcome rate for that band rather than a single blended rate for the whole book. A blended rate produces a number that is right on average and wrong in every month, which is the pattern that makes renewal forecasts feel unpredictable.
Forecast downgrades as their own line. Seat and tier reductions have different causes than full logo churn and need to be predicted separately, or the contraction they carry never appears in the forecast.
How do you tie the renewal forecast back to the total revenue forecast?
Report renewal and new business as separate lines that sum to the total, never as a blended figure. Board audiences need to see which engine produced the quarter, and operators need to know which one to fix.Track accuracy separately as well. Keep a renewal forecast accuracy history alongside your new and expansion history, using the same snapshot discipline: archive the number at the start of the period, leave it unadjusted, and score it at close. If your renewal forecast is consistently high, check whether the risk bands are stale. If it is consistently low, check whether contraction is being double-counted against churn.
The advantage renewal forecasting has over new business forecasting is the known date. You know twelve months in advance which contracts come up and when. That advantage only pays off if the health signals feeding the model are real ones, drawn from support and usage data rather than from a sentiment field someone fills in the week before the renewal call. For how the renewal line fits into the full revenue picture, see how to forecast revenue.
Frequently Asked Questions
Should renewals be forecast separately from new business?
Yes. Renewal revenue has different drivers, a different time horizon, and different early signals than new and expansion business. Blending them into one number hides which side of the business is actually moving and makes both forecasts harder to diagnose.
What is the earliest signal that a customer will churn?
Support case volume at either extreme. A customer with no support cases at all is at risk, because nothing is being used. A customer with seven or more cases in a year is also at risk. Three to five cases, usually tier two or three rather than severe, correlates with an engaged and healthier account.
Why does an account with no support tickets look risky?
Because it usually means nobody is in the product. Silence reads like satisfaction and is more often disengagement. The absence of a signal is a signal, in renewals the same way it is in new business deals.
What does the ARR waterfall need to include?
Beginning ARR, churned customer ARR, churned product ARR, product decrease ARR, new customer ARR, new product ARR, increased product ARR, and ending ARR. Beginning ARR for each month equals the prior month's ending ARR, so the whole sequence reconciles.
How far ahead should a renewal forecast run?
Twelve months on a rolling basis, because contract dates are known that far out. That known-date advantage is what makes renewal forecasting more tractable than new business forecasting, provided the health signals feeding it are real.
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