Credeity publishes its methodology so lenders can evaluate the analysis the way they would any underwriting input. This page describes how reports are produced, how findings are computed, and the limits of what can be inferred from the underlying data. The same material appears in every issued report.
What the analysis is
The Underwriting Intelligence Report is an independent third-party review of observed payment and receivables behavior for healthcare practices. The analysis is lender-initiated and borrower-authorized. Credeity does not make credit decisions; the report is one input among several in the lender's underwriting process.
Data sources
Findings derive from practice-furnished accounting system exports of vendor obligations and payment history (QuickBooks or equivalent), borrower-provided bank statements for the reporting period used for Tier 1 reconciliation where available, and accounts receivable aging reports as of a stated cutoff date. Reports are point-in-time artifacts tied to a stated reporting window and report date.
Deterministic scoring
The Payment Discipline Index is a bounded 0 to 100 score derived from outbound payment behavior. Scoring is deterministic: the same validated payment set always produces the same score. Tier assignment drives how late payments affect the PDI. A payment counts as on-time when days late fall within defined timing buckets. Interpretation bands are a reporting convention: Strong 90 and above, Monitored 85 to 89, Elevated 75 to 84, Acute below 75.
Risk verdict and analyst notes
Each report carries a single canonical risk verdict derived from the PDI, critical-obligation flag context, and severe lateness signals. When multiple thresholds are crossed, the verdict reflects the most actionable proximate cause. Analyst notes are generated by deterministic rules over the report data, not by discretionary authoring. They surface in severity bands with caps on how many notes appear, ordered by severity. Underwriting flags are produced by the same rules engine from the same validated payment set.
Quality controls
The ingestion layer emits structured validation warnings covering duplicates, date-range issues, future dates, ambiguous date formats, zero amounts, low row counts, and missing critical fields, so operators can trace upstream data issues. Deduplication and exclusion rules are deterministic and applied before any scoring runs. A minimum-row gate is evaluated after deduplication to flag thin datasets. Edge cases receive documented defaults rather than silent failure, with internal logging of assumptions.
Independence and verification
Bank statements are reviewed as an independent verification layer: borrower-provided accounting data is confirmed against actual bank debits before the report is issued. Categories that cannot be matched to a bank debit are labeled unverified in the report rather than assumed. External validation items in the Comprehensive tier (UCC lien status, IRS Form 941 transcript, professional license) are drawn from lender-obtained or third-party sources and summarized in a standardized panel.
Known limitations
Revenue and collections are not independently verified unless stated. The analysis covers the stated reporting window only; behavior after the window is outside scope. Personal financial analysis of the principal guarantor is outside the scope of the report. Latency between practice accounting activity and report issuance equals the practice's export cadence plus validation and aggregation time. Each report states its own limitations inline.
Model governance
The PDI scoring methodology is deterministic and versioned. Material methodology changes are documented, and reports state the conventions in force at issuance. The PDI is a behavioral signal intended for use alongside, not in place of, the lender's existing diligence.