Research Methods

SBA Data Methodology

Current dataset through June 30, 2026. Retrieved and verified August 8, 2026.

1. Primary source

The core source is the U.S. Small Business Administration's public7(a) and 504 FOIA loan-level dataset and data dictionary. The current build contains 281 active lender profiles using records available through June 30, 2026.

Supplemental institution information may come from FDIC BankFind and public institution data, FFIEC Community Reinvestment Act disclosures, NCUA call reports, the U.S. Treasury CDFI Fund certified institution list, and official lender websites. Supplemental fields are identified on lender pages when they appear.

2. Coverage and inclusion

• The directory covers SBA 7(a), SBA 504 Certified Development Company, and 504 third-party lender records where available.
• The 7(a) source file begins in 2020. The 504 source file begins in 2010. Combined figures therefore can cover different starting periods by program.
• A public profile requires at least one lender-program record with 50 or more approvals across FY2025 and FY2026 to date. Records for the same normalized institution are then merged into one profile.
• Loan-size, rate, term, industry, state, and business-age distributions use funded records available in the relevant source file unless a page states a narrower period.
• Recent counts combine FY2025 with the available portion of FY2026. FY2026 is incomplete and is not used as a completed annual trend period.
• SBA fiscal years run from October 1 through September 30.

3. Lender identity and program handling

We standardize display labels and create a stable page identifier for each lender. Program roles remain separate when the records distinguish a 7(a) lender, a Certified Development Company, and the first-mortgage third-party lender in a 504 project. A merger, acquisition, name change, or inconsistent source label can split or combine historical activity imperfectly. Report suspected identity problems through the Corrections Policy.

4. Metric definitions

5. Ranking rules

Fastest lenders

Ranked by median approval-to-first-disbursement days. A lender must have at least 100 usable disbursement observations and at least 100 recent approvals. Ties are broken by observation count.

Startup activity

Ranked by the share of funded 7(a) records with a known business-age value that are classified as a startup or a new business of two years or less. The share uses the available 7(a) source window. A lender must also have at least 100 recent 7(a) approvals to demonstrate current activity. Ties are broken by recent 7(a) volume.

Approval growth

Ranked by the absolute number of annual approvals added from FY2023 to FY2025. A lender must have at least 100 FY2023 approvals. FY2026 is excluded because it is incomplete.

Industry rankings

Ranked by funded loan count in the stated NAICS grouping across the available source files. Restaurant and food-service research uses sector 72. Trucking and transportation research uses sectors 48 and 49. Figures labeled as a lender's median loan, rate, or startup share may describe the lender's broader SBA portfolio rather than only that industry. Each page states this limitation.

Business acquisition shortlist

The public FOIA file does not contain a reliable loan-purpose or change-of-ownership field. We therefore do not claim to rank acquisition volume. The shortlist screens for recent 7(a) volume, median approval size, approval range, and usable disbursement observations, then tells readers to verify acquisition experience directly.

6. Records excluded from a calculation

A metric excludes records when its required field is missing, invalid, or unusable. Funding-speed calculations exclude records without both approval and first-disbursement dates and dates that do not create a usable nonnegative interval. A missing metric stays blank. It is not converted to zero and does not receive a favorable ranking.

7. Known limitations

• Federal records describe past activity and do not publish lender decline rates or current credit policy.
• The dataset does not include borrower revenue, complete application dates, or a dependable acquisition-purpose field.
• The 7(a) file begins in 2020 while the 504 file begins in 2010, so combined historical counts do not always share one start date.
• First disbursement can occur well after approval for reasons outside a lender's control, including construction, appraisal, closing, and borrower readiness.
• Some institutions report disbursements at month end, which can make the calculated gap appear longer.
• Rates are initial recorded rates, not live pricing, APR, fees, or total cost.
• SBA 504 records can represent different parties and portions of the same project.
• Reporting corrections, lender name changes, and delayed source updates may alter historical results.

8. Update schedule

We target a refresh after material SBA dataset updates, generally at least quarterly while the directory is active. The cutoff and verification date appear on research pages. A major method change will be documented here and may cause rankings to move even when underlying lending activity has not changed.

9. Reproducing or challenging a result

Start with the SBA dataset and data dictionary linked above. Then send the page URL, lender name, disputed metric, source record or file version, and your calculation tolatamresourcegroupllc@gmail.com. We follow the Corrections Policy and do not accept payment to alter a ranking.