The three bases teams actually use
| Denominator | What it answers | Where it breaks |
|---|---|---|
| Revenue or logos at the start of the period | How much of the base did I hold | Ignores anything sold and lost inside the period |
| Average of start and end balances | Smooths a base that grew sharply mid-period | Blends acquisition into a retention metric |
| Revenue available to renew in the period | How well did I convert the renewals I had | Hides losses from accounts not up for renewal |
Rules that keep the base stable
Freeze the base on day one and refuse to restate it. Retroactive edits to starting ARR, whether from a corrected contract value or a late booked amendment, quietly rewrite every historical churn rate you have already published.
Count each account once. Parent and child records, multiple subsidiaries under a single master agreement, and duplicate CRM entries all inflate the denominator and understate churn. Decide which level the contract lives at, then apply that level everywhere.
Hold currency constant. Retention measured on ARR converted at spot rates mixes exchange rate movement into a customer behavior metric, and a strong dollar quarter will read as a churn problem.
Separate logos from dollars. A logo denominator counts accounts, a revenue denominator counts ARR, and losing four small accounts is a very different event from losing one large one.
Why the base matters for forecasting
A model that projects retention forward inherits whatever denominator produced its training history. Feed it a renewal rate series, label the output gross revenue retention, and the model will overstate the surviving base every period, because renewal rate never saw the mid-contract cancellations. The gap compounds across a multi-year plan. Anyone building a revenue forecast on retention history should confirm which base each period used before the series goes near a model.
The practical test is simple. If two people in the company can produce different churn rates for the same quarter and both are correct, the denominator is not documented well enough. Write the base into the metric definition, print it in the report next to the rate, and keep it identical across net revenue retention, gross retention, and logo churn so the three reconcile to the same ARR movement.
Frequently Asked Questions
Should customers acquired during the period count in the churn denominator?
No. A customer who signs in week three of the quarter and cancels in week ten belongs in a new logo quality review, not in the retention base for that quarter. Including mid-period acquisitions makes churn look worse in fast growing quarters and better in flat ones, which inverts the signal you wanted.
What denominator does gross revenue retention use?
Recurring revenue on the books at the start of the period, held constant. Every contraction and cancellation during the period counts against that frozen base, including losses from accounts that were nowhere near a renewal date.
Why do churn rate and renewal rate disagree at the same company?
They use different bases. Renewal rate divides by revenue available to renew in the window. Churn rate against the full base includes accounts that had no renewal event at all, so mid-contract cancellations show up in one number and vanish from the other.
Is an average of beginning and ending revenue an acceptable base?
It smooths a base that grew sharply during the period, which is why finance teams reach for it. The cost is that acquisition leaks into a retention metric, since the ending balance includes revenue you sold rather than retained. Use it only when you disclose it.
Put these metrics to work
ORM builds custom revenue forecast models that turn concepts like churn rate denominator into prescriptive action for your team.
Schedule a Demo