How the Bank Fee Index works

Data Sources

We start with every regulated U.S. bank and credit union

Bank Fee Index draws its institution universe from two authoritative federal databases: the FDIC's BankFind Suite (which tracks every FDIC-insured bank, thrift, and savings institution) and the NCUA's Research & Data portal (which covers all federally chartered credit unions). Together, these sources provide accurate legal names, charter classifications, asset sizes, physical locations, and primary website URLs for roughly 8,750 active institutions.

We do not use purchased data lists, scraped directories, or self-reported feeds. Every institution in our index is traceable to a federal regulator record with a published institution ID. This is the foundation of our data quality commitment: our institution universe is authoritative before the first fee is collected.

As of the most recent index update, 1,183 institutions have verified fee schedules in the Bank Fee Index, across 50 states. Coverage is skewed toward institutions with assets above $100 million, where fee schedules are most consistently published online. Institutions below $50 million in assets are included where fee schedules are publicly discoverable.

Collection Process

Automated collection runs on a rolling calendar

Automated collection runs on a rolling calendar and stores the document, its URL and the date collected. Every fee in the index points back to that stored document, so a figure can always be checked against the schedule it came from.

Finding the schedule comes first. Most institutions publish it at a predictable address (a fee schedule PDF, or a disclosures page); where they do not, we search the institution's own website. Large regional banks publish reliably; community banks and credit unions are more variable, and coverage is weakest where the schedule is not published online at all.

We read HTML pages, plain-text documents and PDFs with selectable text. Scanned, image-only PDFs are set aside for separate handling rather than mixed into the general queue.

Institutions whose fees change often are rechecked more often. Every schedule is rechecked at least quarterly. When a stored document has not changed since the last visit, its fees are carried forward rather than re-read.

Reading the fees

Only what the document says, and a person checks the rest

From the text of each schedule we record the fee name, the amount, and the conditions attached to it — waivers, tiers, whether it is charged per item or per month.

Fees the software is not sure about are held for a person to check. Only fees that are clearly stated, or that a reviewer has confirmed, appear in the public index.

We do not infer or estimate fees. If an amount is not written in the document, none is recorded. This is the primary safeguard against invented fee data.

Each fee is stored with its source document, the date collected and its review status, so there is a full trail from the published schedule to the index entry.

Categorization

Standardized fee categories make institutions comparable

Raw fee names vary substantially across institutions. "Monthly service charge," "account maintenance fee," and "checking maintenance" typically refer to the same economic product. Comparison is only possible after normalization.

Bank Fee Index maps every raw fee name to a standard category — 57 categories currently carry verified data — organized into fee families such as account maintenance, overdraft and NSF, wire transfers, ATM and card, check services, and account services. Each category has a canonical name and a maintained list of known aliases.

Categorization is automatic when a raw fee name matches a known alias. Names that do not match are held for a person to assign, and the alias list grows as new naming patterns appear.

A small set of spotlight categories (monthly maintenance, overdraft, NSF, non-network ATM, foreign transaction, domestic outgoing wire) appears at high rates across all institution types and anchors the public index; the full list of categories is on the Bank Fee Index page.

Checks before publication

Uncertain fees are held; outliers are looked at by a person

Before any fee enters the published index, it passes two checks.

First, certainty: fees the software is not sure about are held for a person to check and are excluded from public benchmarks until confirmed.

Second, outliers: fees far outside the rest of their category — an ATM fee of $300 when the category median is $3.00 — are flagged and reviewed. Flagged fees are confirmed, corrected, or excluded.

Every category and every institution carries a plain status. Verified: 10 or more checked fees, benchmarked publicly. Under review: fees collected but still being checked. Too few to benchmark: fewer than 10 verified fees, shown but not used for medians.

Coverage and Limitations

What our data covers — and what it does not

Bank Fee Index tracks published fee schedules, not actual fee revenue or transaction-level data. A published fee of $35 does not mean a given institution collected $35 for every overdraft — waiver programs, promotional rates, and negotiated terms affect realized fees. Our data reflects disclosed rates, which are the standard of comparison for regulatory purposes and consumer research.

Our coverage is strongest for retail deposit account fees (maintenance, overdraft, NSF, wire, ATM) and weakest for business account fees, loan fees, and investment-account fees. Fee schedules for these product types are less consistently published in machine-readable formats.

Geographic coverage is reasonably uniform at the state level but skewed toward states with higher institution density (Texas, California, Illinois, Ohio, New York). Fed District 4 (Cleveland), District 7 (Chicago), and District 11 (Dallas) have the strongest coverage. District 10 (Kansas City) and District 12 (San Francisco, excluding California) have the largest gaps relative to institution population.

Source freshness varies by institution and schedule. The national index represents a rolling snapshot of fee schedules collected over the trailing 120 days. State-level indexes use a 90-day window. Fees older than these thresholds are excluded from the live index to prevent stale data from distorting benchmarks.