Methodology · Last updated September 30, 2026

How the franchise SBA default rates are built

Every figure on this site derives from the U.S. Small Business Administration's public 7(a) FOIA loan file (data.sba.gov). A brand's default rate is the share of its SBA 7(a) loans approved FY2010–FY2019 that were charged off, with loan statuses as of June 30, 2026, measured across 415 franchise brands and 26,660 resolved loans.

Where does the data come from?

Every number on this site traces back to one federal source: the SBA 7(a) FOIA loan file, released by the U.S. Small Business Administration at data.sba.gov under the Freedom of Information Act (dataset family "7(a) & 504 FOIA"). It is not a survey, an estimate, or a franchisor's own disclosure. It is the government's ledger of every 7(a) loan it guaranteed, reported by the lender that wrote it and tracked by the SBA until the loan is repaid or written off. A franchisor can choose what goes into Item 19. Nobody chooses what goes into this file.

Each record is one loan, carrying exactly the fields a credit analyst needs: the approval date, the gross approval amount, the franchise code and brand name where the borrower is a franchisee, the NAICS industry code recorded at origination, and the loan's status today (paid in full, charged off, cancelled, or outstanding). Because the status is set by the lender and the SBA, not by the brand, a charge-off here is a loss that a bank and the federal government actually took.

The SBA refreshes the file quarterly. This site runs on the release with loan statuses as of June 30, 2026; that date is printed on every page, and every figure is recomputed from the raw file at each refresh. No franchisor, broker, or consultant supplies data, reviews it, or pays for placement.

That is the difference a Lender Report ($29) or FDD Teardown ($500) buys you: the same public record the SBA and its lenders use, cleaned, matched to the brand, and computed the way an underwriter would, rather than the figures a franchisor chooses to present in Item 19.

Which loans count, and why is the cohort FY2010–FY2019?

Rates use loans approved in federal fiscal years 2010 through 2019, the most recent cohort old enough to have been tested through a full loan term. The denominator is resolved loans — paid in full plus charged off. Cancelled and never-disbursed loans are excluded upstream because no money went out. Loans approved from FY2020 onward are reported as counts only (loans and early charge-offs), never as a rate. Most of those loans have not reached maturity. A default rate on an unresolved cohort is censored: the numerator can only grow. Any percentage published today would understate the eventual rate and reward the youngest, least-tested brands. That is the censoring problem. Publishing a flattering early number would be easy. This site waits for the cohort to mature instead.

Most SBA default figures published online use loans approved since 2020 — loans that have had six years or less to fail — so those rates are floors, not outcomes. Every rate on this site is computed from the matured FY2010–FY2019 cohort: loans that have been paid in full or charged off, so the result is settled. Loans approved since 2020 are shown as counts so early charge-offs are visible without being mistaken for a rate.

How were the 10.2%, 8.0%, and 7.8% benchmarks computed?

All three come from the same FY2010–FY2019 cohort and the same resolved-loan denominator. The all-franchise 7(a) charge-off rate of 10.2% pools every loan carrying a franchise code. The all-borrower rate of 8.0% pools every resolved 7(a) loan in the cohort. The non-franchised rate of 7.8% pools resolved loans with no franchise code. The industry average on brand and industry pages pools every resolved 7(a) loan in that brand's primary industry, franchised and independent alike. A brand is therefore compared with all SBA borrowers in its line of business, not with the whole economy. Industries with fewer than 100 resolved loans show no average.

Does the franchise gap replicate in other datasets?

Yes. In the 1990s 7(a) cohort, franchised borrowers charged off at 15.5% versus 11.9% for non-franchised borrowers. In the 2010s SBA 504 program — a different loan product with different lenders and collateral — franchised borrowers charged off at 4.0% versus 2.7% for non-franchised borrowers. Different decades and a different program show the same direction: franchised borrowers default more often than comparable independents.

How are brand names normalized?

The SBA records franchise names inconsistently ("WINDOW GENIE", "Window Genie", "9ROUND; 9round kickboxing"). We uppercase each name, drop punctuation and parenthetical text, key on the first segment before a semicolon, and merge rows that share a key: loan and default counts are summed, rates recomputed, and average loan sizes weighted by loan count. Display names use the cleanest variant. Brands whose merged sample is under five loans are not published as pages.

What are the known limitations?

Any dataset presented as flawless is being sold to you. These are the limits of this one, stated the way an underwriter would want them stated:

• SBA borrowers skew undercapitalized relative to all-cash buyers, so these rates can overstate risk for well-funded operators.
• Franchise tagging covers only roughly 8–12% of 7(a) loans. That share is not an outside estimate: it is the fraction of records in the FOIA file that carry a franchise name, counted from the file itself at each quarterly refresh. Brands with few SBA-financed franchisees will have small samples.
• Small samples move a lot: at 20 loans, one default is five percentage points. Pages state the sample size and a caveat wherever it applies.
• Industry codes are self-reported at origination and are not audited.
• Charge-off is a lending outcome, not a business-closure count; some failed units never had SBA loans and some charged-off borrowers kept operating.

How is this site funded?

How this site is funded. Franchise Failure Rates takes no money from franchisors, franchise brokers, or franchise consultants. We do not sell leads, claimed listings, or franchisor reports. Revenue comes from reports and data sold to prospective franchisees, their attorneys, and lenders.

How are corrections handled?

This is independent research from public U.S. government data. We are not affiliated with any franchisor, broker, or franchise consultant, and nothing here is legal, financial, or investment advice. The standard is the same one we would hold a franchisor to: a figure stands only while the record supports it. Franchisors and franchisees who believe a figure is wrong can email hello@franchisefailurerates.com with documentation — a loan-level extract, a franchise-code mismatch, or evidence of a merge error. We review within 5 business days, correct the data file, rebuild every page from the corrected file, and note the change here in the open.

How can the numbers be reproduced?

Nothing here asks to be taken on trust. Every figure derives from the public files at data.sba.gov — dataset "7(a) & 504 FOIA" — and anyone with a spreadsheet can check it. Filter to approval fiscal years 2010–2019, keep loans with a franchise code, drop cancelled loans, group by normalized franchise name, and divide charged-off by resolved loans. The site's cleaned brand table is published as CSV and JSON; a machine-readable index of pages is at /llms.txt.