Cohort reconciliation formulas

The count reconciliation difference equals observed ending customers minus expected ending customers. A nonzero amount is a data-quality prompt, not an automatic adjustment to the retention rate.

Worked annual customer-cohort example

The default business opens with 10,000 active customers and loses 1,800 of those logos. The retained opening cohort is 8,200 and annual logo retention is 82 percent. The business also acquires 2,500 new customers and reactivates 300 former customers, bringing expected ending customers to the observed 11,000.

Net customer growth is ten percent even though retention is 82 percent because acquisition exceeds churn. New and reactivated inflows represent 25.45 percent of ending logos. Retention is three percentage points below the prior entered 85 percent. Growth and retention therefore need separate labels.

The opening cohort begins with $1,200,000 recurring revenue. Churn removes $150,000 and contraction removes $80,000, producing 80.83-percent gross revenue retention. Expansion adds $220,000, so ending cohort revenue is $1,190,000 and net revenue retention is 99.17 percent. Revenue from new customers is deliberately excluded.

Choose one customer unit

Do not switch between users, seats, accounts, and contracts to improve the number. Maintain a durable identifier, deduplicate, and record policy changes.

Lock the opening cohort before measuring

Create a list of customers active at the opening timestamp. Follow those same identifiers to the ending timestamp. New logos cannot repair logo retention because they were never at risk in the opening cohort. Reactivations should be shown separately unless the published policy explicitly treats them differently.

Exclude test, employee, fraudulent, and duplicate accounts under a written policy applied before results are known. Document acquired or divested customer bases, currency changes, product migrations, and threshold changes.

Define the churn event and effective date

A cancellation request, service end, contract expiration, nonrenewal, failed payment, account deletion, or inactivity threshold can mark churn. Grace periods and pauses complicate timing. A customer may cancel today but remain entitled to service through month-end.

Use a date aligned with the revenue and service policy. Separate voluntary churn, involuntary payment failure, customer business failure, product migration, and administrative cleanup so teams can act on causes.

Monthly and annual rates do not scale linearly

A monthly retention rate cannot be multiplied by twelve. Under a constant survival assumption, annual retention is the monthly rate compounded twelve times. Real businesses may renew annually, churn during onboarding, or show seasonal behavior, making direct cohort measurement preferable.

Compare equal-duration periods and cohort ages. A rolling twelve-month rate and a calendar-year cohort rate can include different customers and events even when both say “annual.”

Logo, gross revenue, and net revenue retention

Logo retention weights every customer equally. Gross revenue retention asks how much opening-cohort revenue remains after churn and contraction before expansion. Net revenue retention then adds expansion from the surviving opening cohort and can exceed 100 percent.

A high net rate can hide logo churn if a small number of large customers expand. A high logo rate can coexist with revenue loss if large customers contract. Review both plus customer concentration and cohort size.

Expansion must belong to the opening cohort

Upgrades, seats, usage, cross-sell, and price changes can expand recurring revenue. Count only amounts from customers present at the start. Revenue from new logos belongs to acquisition, even if it arrives through an existing sales channel.

Separate contractual price increase from additional consumption when analyzing product health. Track downgrades and concessions as contraction. Reconcile every customer movement to the ending cohort revenue.

Reactivation deserves its own movement

A former customer who returns is valuable, but treating the logo as continuously retained erases the churn event and overstates survival. Record churn and later reactivation separately, with a rule for short pauses and failed-payment recovery.

Measure reactivation cost, time away, retained revenue, and repeat churn. A discount-driven win-back can grow ending customers while producing weak contribution or training customers to cancel.

Diagnose retention by customer journey

Segment activation, first value, onboarding completion, support contacts, product adoption, contract milestone, renewal, and cancellation reason. Compare cohorts by acquisition channel, plan, geography, customer size, sales motion, initial offer, and product version.

Use sufficient sample size and preserve denominators. A high percentage from a tiny segment may not be decision-ready. Combine quantitative movement with customer interviews and support evidence.

Measure interventions against a credible comparison

A retention campaign may target customers already most likely to stay, making the participants look better even if the campaign has no effect. Define eligibility before outreach and use randomized holdouts, phased rollouts, matched cohorts, or another defensible comparison when practical. Measure incremental retention and contribution, not only open rates or accepted offers.

Include discount, service, support, product, and operational cost. A save offer that retains a customer at negative contribution can increase logo retention while reducing value. Track whether the customer remains after the incentive ends and whether repeated save offers change cancellation behavior.

Predefine the observation window, churn event, reactivation treatment, and minimum sample. Preserve customers in the group originally assigned even when they do not engage, so the test does not reward self-selection. Report confidence and practical dollar impact beside percentage-point movement.

Forecast the installed base with movement rates

Start with opening customers, apply a cohort-age retention curve, add expected acquisition and reactivation, and reconcile the next ending balance. New customers become a separate cohort with their own onboarding risk; do not immediately assign them the mature-base retention rate.

Build monthly movement schedules for staffing, support load, billing, revenue, cash, and product capacity. Stress acquisition, early churn, mature churn, contraction, expansion, and price separately. A flat ending-customer target can hide excessive replacement of recently churned logos.

Metric disclosure should survive replication

SEC KPI guidance emphasizes definition, usefulness, calculation, and changes in method. Current SEC filings demonstrate that public companies define ARR cohorts, minimum thresholds, subscription units, and exclusions differently.

Store the opening cohort and movement file so another analyst can reproduce the result. Explain policy changes and recast prior periods when appropriate rather than silently improving comparability.

Retention is not recognized revenue or taxable income

Retention metrics describe customers or recurring revenue under a management policy. Revenue recognition, deferred revenue, refunds, bad debt, sales tax, and taxable income follow accounting and tax rules.

Retention close checklist

1. Define the customer unit.
2. Freeze the opening cohort.
3. Define churn and effective date.
4. Separate new and reactivated logos.
5. Reconcile observed ending customers.
6. Map starting cohort revenue.
7. Separate churn and contraction.
8. Map expansion to opening logos.
9. Reconcile ending cohort revenue.
10. Compare equal period lengths.
11. Segment causes with stable rules.
12. Preserve a reproducible movement file.

Frequently asked questions

References

• U.S. Securities and Exchange Commission, KPI and metric disclosure guidance
• 2026 SEC filing illustrating ARR net retention and gross retention definitions
• 2025 SEC filing illustrating annual small-business retention disclosure