The Fixed-Window Payment Recovery Review Framework

FIXED-WINDOW RECOVERY REVIEW — original Growthcraft synthesis
DECISION: {{question, proposed change, owner, review date}}
SCOPE: {{provider, renewal vs first invoice, currency, failure cohort, extraction cutoff}}
WINDOW: {{elapsed days, chosen before reading outcomes}}
IDENTITY: {{invoice key, first qualifying failure, full-settlement definition}}
EXCLUSIONS: {{partial payments, write-offs, disputes, voids, refunds, missing histories}}

1. DEFINE — billing + finance
Do numerator and denominator share invoice grain, amount boundary and currency?
Are the amounts the outstanding balances frozen at first failure?
Have cohort-entry rules been frozen independently of later outcomes?
Evidence / discrepancy / accountable owner:
2. RECONCILE — analytics
Mature count M = within-window R + late L + unpaid U.
Mature balance B = within-window C + late D + unpaid E.
Immature count and amount (excluded, even if already paid):
Source-completeness evidence / omitted rows / extraction lag:
3. INTERPRET — growth operations
Invoice rate R/M; amount-weighted rate C/B:
Why those rates differ / concentration / still-unknown causes:
Policy mix / attempt execution evidence / comparison population:
4. ACT — named approver
HOLD if history, eligibility or permission is missing.
INVESTIGATE a specific discrepancy with owner and due date.
An intervention needs a separate approved design and customer guardrails.
Never infer incremental uplift, profit or saved customers from descriptive recovery.
Smallest missing evidence that would change the decision:
Cost boundary, complaints guardrail and rollback owner:

Use before changing a retry or messaging policy

This original framework is for subscription growth leads, billing engineers and analysts comparing recovery reports. It separates invoice eligibility, reconciliation, interpretation and action. Use it when one period appears to recover less, a provider switch is proposed, or retry activity is being presented as recovered revenue. Do not use it to collect debts, determine payment eligibility, automate customer contact or calculate net profit.

Prerequisites and owners

Billing owns a unique invoice key and first-failure history. Finance defines the outstanding balance, treatment of tax, credits and refunds, and currency policy. Analytics freezes the window and extraction cutoff. Growth owns the question, not an invented explanation. A named commercial owner approves any change. Supply anonymised invoice balances and elapsed times; never paste card details, emails or secret payment links.

Four gates, not an averaged score

1. Define: separate renewals from trial-to-paid invoices. Choose a fixed observation window and cohort-entry period. Hold if the definition changed halfway through.
2. Reconcile: one failed invoice enters once, no matter how often it is retried. Keep immature invoices outside both numerator and denominator. Reconcile mature counts and frozen balances against approved exports.
3. Interpret: show both invoice-count and amount-weighted recovery. Identify whether a few large invoices dominate. Distinguish late recovery from unpaid-at-cutoff; neither belongs in the fixed-window numerator.
4. Act: choose hold or a bounded investigation before recommending a policy change. Arithmetic does not establish customer need, causal effect or permission to contact someone.

Worked review: a misleading headline

Synthetic example, EUR, 30-day window. A/B/C/D are each 35 days old with balances 100/200/100/600. A settles on day 3, B on day 20, C is still unpaid and D settles on day 32. E is only 10 days old, balance 100, settled on day 2. Mature balances total 1,000; within-window settlement is 300. Count recovery is 2/4 = 50%; amount recovery is 300/1,000 = 30%. E is excluded despite its payment; D is late despite eventually settling.

The review should not call the programme ineffective or claim 900 recovered within 30 days. Ask billing to explain the large late invoice and analytics to verify cohort completeness. If no export reconciliation is supplied, mark HOLD. If completeness is proven but the reason for lateness is absent, assign a bounded investigation. Thirty days is the example's chosen window, not a universal target or a claim about every provider.

Failure modes and limits

Repeated events inflate invoices if identity is wrong. Segmenting on the final payment method selects successful outcomes. Changing amounts after refunds breaks the frozen denominator. Excluding immature unpaid invoices but including immature paid invoices creates a winner-selection bias. A voided invoice is not a payment. This full-invoice convention cannot represent partial settlement or net cash; use a payment-allocation model for those questions. No benchmark is needed to enforce exact count and balance reconciliation.

Sources and boundaries

Primary documentation checked 24 September 2026: Stripe payment retries explains that scheduled attempts and executed charges can differ; Paddle Payment Recovery describes a default 30-day dunning period. Neither source validates this original cross-provider measurement convention or guarantees a recovery rate. Dates of these living pages are not stated. Check your actual account configuration.