Methodology

Friction scores summarize public CFPB Consumer Complaint Database activity relative to institution size (FDIC assets), after mapping CFPB company names to FDIC CERT via a curated bridge table.

Score version: friction_v1

1. Count complaints with Date received in the trailing 12 months, by CFPB Company name.
2. Map company names → FDIC CERT using a curated bridge (manual for large banks; assisted unique name matches among institutions with ≥ $10B assets).
3. Sum complaints for each CERT. Compute rate = complaints ÷ assets in billions of dollars (FDIC ASSET / 1,000,000).
4. Convert rate to a 0–100 friction index: 100 × log10(1 + rate) / log10(1 + p95_rate), where p95_rate is the 95th percentile rate among scored banks. Higher = more complaint friction per $1B assets.

What we will not claim

• Banks with ≤ $10B assets: CFPB does not publish referred complaints in the public database. We show No public CFPB complaint score — never “community bank verified.”
• Large banks without a bridge match show unmatched — not a clean bill of health.
• Scores are not investment, legal, or consumer advice.
• We are not affiliated with FDIC, CFPB, or any bank.

Data sources

• FDIC BankFind — institutions, locations, assets
• CFPB Consumer Complaint Database (full CSV dump, aggregated locally)
• Federal Reserve E-Payments Routing Directory (public search)

Version id lives in friction_scores.score_version. About