Contraction MRR vs Churned MRR: How to Classify Every Loss

What Is the Difference Between Contraction MRR and Churned MRR?

Churned MRR is revenue lost when a customer leaves entirely, and contraction MRR is revenue lost when a customer stays and pays less. The customer relationship is the dividing line.

A company that cancels its contract and stops paying produces churn. A company that renews at 250 seats instead of 400 produces contraction. Both reduce the base by real dollars, and both land in the same gross retention calculation, so a team that only watches the headline number cannot tell which one is happening.

They are also different problems. Churn is a relationship and value failure that has already concluded. Contraction is a value failure in progress, with a customer still on the other end of the phone. Merging them into one loss line eliminates the distinction that would tell you whether to run a win-back play or an adoption play.

How Should Each Type of Loss Be Classified?

Sort every loss event by whether the account survived and whether the reduction was full or partial. Four categories cover almost every case. At ORM the monthly retention waterfall uses this structure directly. Beginning ARR, churned customer ARR, churned product ARR, product decrease ARR on the contraction side, then new customer ARR, new product ARR, and increased product ARR on the expansion side, ending at ending ARR where beginning ARR for each month equals the prior month's ending ARR. That reconciling waterfall is what makes retention explainable rather than approximate, because every dollar of movement belongs to a named line.

Why Does the Split Change What You Do About It?

Contraction and churn have different causes, different owners, and different windows for action. Running one play against both wastes the window.

Contraction usually traces to usage and value delivered against spend. Seats bought and never provisioned, a department that never onboarded, a use case that stalled after the initial rollout. The response is adoption work, and it can start the moment the downgrade request arrives, well before the renewal paperwork.

Churn traces to something larger. A champion leaving, a competitive replacement, a budget cut, an acquisition that consolidates vendors. By the time it is recorded, the decision was made weeks earlier. The response is a save play if you catch it in the risk window and a win-back campaign if you do not.

The two also predict differently. Contraction is the earlier signal and it is measurable. An account that keeps contracting is telling you the renewal is at risk, and treating that contraction as a small revenue event rather than a warning is how a churn number arrives with no notice.

How Does Each One Show Up in Retention Metrics?

Both reduce gross revenue retention, and neither is offset by expansion inside GRR. Net revenue retention is where expansion enters and can mask both.

Work a quarter. A base of $8 million loses $240,000 to full cancellations and $160,000 to downgrades, while expansion adds $560,000. Gross retention is $8 million minus $400,000 over $8 million, or 95%. Net revenue retention is 102%, since expansion more than covers the losses.

Report only NRR and the quarter reads as growth. Split the lines and it reads differently. Contraction is 2% of the base, cancellations are 3%, and expansion is doing the work of covering both. That is a business where the base is being replaced from inside itself, which is a very different story from one where GRR sits at 99% and expansion is pure upside.

What Is the Earliest Signal for Each?

Support activity predicts churn, and usage against entitlement predicts contraction. Both are available months before the revenue moves.

Support case volume is one of the most useful churn indicators we see at ORM. Customers with no support cases at all are at risk of churn, and customers with seven or more in the last year are also at risk. The healthy range is three to five cases, usually tier two or three severity, because those accounts are engaged and getting help. Silence reads as safety on a dashboard and it is the opposite.

Contraction signals live in entitlement data. Licensed seats versus active seats, contracted volume versus consumed volume, products purchased versus products with any activity in the last 90 days. A customer using materially less than what they pay for will ask for the difference back at renewal, and the gap is visible long before the renewal date.

How Do You Forecast Contraction and Churn Separately?

Predict them as two distinct events with different drivers, then subtract both from the base in the same waterfall. A single blended churn assumption cannot produce either number reliably.

Contraction is closer to a continuous process. It concentrates at renewal dates but happens throughout the year via amendments, and it scales with the gap between entitlement and usage across the base. Churn is discrete and account-level, driven by health signals and renewal timing.

Model them at the account level and roll up, rather than applying one percentage to the whole base. The rolled-up forecast then feeds the same model that produces your revenue forecast, so the retention line and the new business line are built from one set of assumptions instead of two that get reconciled in a meeting.