Bank Coverage

Which banks borrowers use, how well we handle each, and which new ones are showing up — volume × failure picks the next integration to fix

snapshot

Bank-statement volume by institution

Weekly files, top 8 institutions vs the long tail · institution detection began , so earlier weeks undercount

Institution league table

Top 40 by files in the trailing 180 days · parse fail = file-level hard failures · cashflow pass = truth-parser cashflow-check pass rate (new table, filling in — shown when ≥10 checks)

New institutions

First seen in the last 30 days with meaningful volume (≥30 files) — coverage gaps arriving on our doorstep

Methodology. One row per bank-statement file (files_file, document_type='bank_statement'), joined to institutions_institution via institution_id; internal/test tenants excluded (same list as the Revenue Report). Institution names are normalized to brands with keyword rules (Chase/JPMorgan → Chase, all Wells Fargo variants → Wells Fargo, …) — the long tail keeps its raw name, so minor variants and OCR noise remain distinct there. Institution detection only began ; earlier files have no institution and weekly volumes before then undercount. Weekly buckets are Monday-start Eastern and include weekends (files arrive whenever borrowers upload). Parse fail = file status 3. Cashflow-check pass = share of files whose last-run parser passed the cashflow check (waterfall_cashflowcheckresult — a brand-new table, so most institutions show "—" until it accumulates history). Trailing 180 days · laminr-prod.