High churn: find where customer losses are rising

High churn means customer losses are above a relevant baseline, rather than above a universal percentage. Diagnose it by comparing equal periods, fixing the customer population at each period's start, and finding which segments explain the rise. Choose an intervention after the numbers distinguish a retention problem from a change in customer mix.

1. Freeze the comparison before counting losses

Use two periods with the same duration and the same loss rule. The churn definition covers the denominator; the diagnostic task here is identifying what changed inside that denominator.

Export one row per customer per observation period. Keep customer ID, period, opening eligibility, opening segment, lost-by-cutoff flag, effective loss date and recorded loss reason. Include opening recurring revenue if finance will review the dollars alongside customer counts.

Use the segment recorded at the start. Moving an account into a different segment after cancellation can manufacture an apparent improvement in the original segment. Exclude customers acquired during the period from that period's opening cohort, while retaining them in a separate acquisition report.

For a business without contracts, first document what absence means under its normal purchase cadence. The ledger-based churn workflow handles that decision; a quiet customer should not become a loss merely because the export has no recent invoice.

2. Rebuild the segment table from the export

Suppose a US subscription business compares July and August 2026, both complete calendar months with the same number of days. Each customer's period row has columns A through G: ID, period, segment, opening-eligible flag, lost flag, reason and opening monthly revenue.

In an Excel summary, put segment names in A2:A3, the earlier period label in K1 and the current label in K2. These formulas count the earlier opening cohort and its losses:

B2 =COUNTIFS(Customers!$B:$B,$K$1,Customers!$C:$C,$A2,Customers!$D:$D,1)
C2 =COUNTIFS(Customers!$B:$B,$K$1,Customers!$C:$C,$A2,Customers!$D:$D,1,Customers!$E:$E,1)
D2 =IF(B2=0,"",C2/B2)

For current opening customers in E2 and losses in F2, repeat the first two formulas with $K$2 instead of $K$1. Set G2 to =IF(E2=0,"",F2/E2) and fill down. Microsoft documents COUNTIFS as counting cells that satisfy multiple criteria.

The total rate divides total losses by total opening customers. It does not average the segment percentages. Rates are displayed to one decimal place; keep unrounded values in the workbook.

These are two opening snapshots. New customers and movement between segments can change their mix; a separate customer rollforward would reconcile those movements.

3. Separate worsening rates from customer mix

Overall churn rose by 6.2% minus 3.6%, or 2.6 percentage points. Enterprise retention did not worsen. Small-business retention did, and small businesses also became a larger share of the opening base.

Hold the earlier customer weights constant and apply the current segment rates:

Current churn at earlier mix = Sum of earlier segment customers × Current segment rate / Earlier total customers

=SUMPRODUCT(B2:B3,G2:G3)/SUM(B2:B3)

This bridge changes segment rates first, then mix. State that order: a bridge that changes mix first allocates the interaction differently. Both must reconcile to the observed total change.

4. Trace the deterioration to an actionable reason

Start with small business because its rate changed. Split that cohort by tenure and recorded loss reason, using the same opening segment rule. Look for a specific explanation supported by account records: an onboarding issue, a contract cohort reaching renewal or payment-related losses.

For lost opening monthly revenue by segment, sum column G under the same eligibility and loss conditions. Microsoft's SUMIFS documentation confirms that the sum can require multiple conditions:

=SUMIFS(Customers!$G:$G,Customers!$B:$B,$K$2,Customers!$C:$C,$A2,Customers!$D:$D,1,Customers!$E:$E,1)

Do not call every reason code a cause. Have the account owner validate it against the cancellation record. Keep unknown reasons visible, assign an owner to review them, and compare subsequent outcomes under the same definition. Use the broader churn analysis workflow for the investigation and churn reduction guide when selecting the response.

Where it goes wrong

• Changed cutoff rules. A different grace period moves losses between months without changing customer behavior.
• Duplicate customer rows. Joining invoice lines to customers can count the same loss repeatedly. Confirm customer-period uniqueness before summarizing.
• Mix mistaken for deterioration. The bridge shows that only part of the total increase came from changing segment rates.
• Reasons treated as proof. A cancellation label identifies where to investigate; it does not establish why the customer left.

Covirage is not affiliated with Microsoft.

Diagnose high churn on your own customer file

Upload your customer-status export with opening cohorts, segments and loss reasons. Covirage maps the columns, its tools compute and check comparable losses, and the external AI model explains the change with every figure traced to its rows. Create a free Covirage account and bring a sample of your own export.

See Covirage for sales insights for the supporting workflow.