How Does Churn Affect a Revenue Forecast

A revenue forecast can be right about bookings and still be wrong about revenue, because churn subtracts from the same base that new business adds to. Net ARR is the difference between two independently uncertain numbers, and the uncertainty on a difference is wider than the uncertainty on either input.

The arithmetic that catches people out

Take a plan calling for 100 units of gross new ARR against 80 units of churn and contraction. Net is 20. Now miss by 10% on each side in opposite directions: 90 of new against 88 of churn leaves net at 2. The inputs were close. The answer was not.

That leverage grows as a company matures. Early on, the base is small and churn barely registers against new bookings. As the base compounds, the churn line grows with it, and eventually a majority of the gross new ARR is spent replacing revenue that left. At that point the retention model matters more to the forecast than the pipeline model does.

Three streams, three models

New business, expansion, and renewal behave differently enough that a single model handles none of them well.

New business is driven by pipeline creation, conversion, and cycle length. Expansion is driven by adoption inside existing accounts. Renewal and churn are driven by expiry schedules, usage, and account health, and much of the outcome is already determined by contracts signed one to three years ago.

ORM's read is that forecast accuracy on new plus expansion, excluding renewal, usually lands around 90%, and that getting there takes significant manual effort and does not stay current as conditions change. ORM targets 95% without manual adjustments, holding from day 1 through day 90 of the quarter and updating as the quarter progresses. Renewal is a separate estimation problem with its own data, which is why it belongs in its own model rather than as a plug.

Retention assumptions go stale the same way pipeline assumptions do

ORM's view on why forecasts miss applies directly here. The most common failure is that something in the business or the market changed while the forecast still runs on old assumptions. Retention inputs are unusually prone to this, because a churn rate is an average over a cohort mix that keeps shifting. Move upmarket and the blended rate improves for reasons that have nothing to do with the product. Land a large cohort of SMB accounts in one quarter and it degrades a year later.

Rate based retention forecasts also ignore timing entirely. Applying an annual retention rate evenly across twelve months assumes contracts expire evenly, and they almost never do.

What actually goes into the model

Build the retention side from the schedule rather than the rate. Available to renew by month for the next four to six quarters. A separate mid-contract loss rate for the portion of the base with no expiry event in the window. Contraction modeled apart from full cancellation, because they respond to different things. Segment level rates instead of one company average.

Then reconcile the streams to the same ARR waterfall the retention report uses, so that sales forecasting output and retention output describe one base rather than two. Comparing the result to plan line by line is what turns a miss into a diagnosis, and it is the difference between knowing the number was wrong and knowing which assumption broke. The mechanics of assembling that view are covered in how to forecast revenue, and the measurement discipline behind it in forecast accuracy.