Data notebook · October 5, 2026 snapshot

From source rows to one count per UEI

A row, a registration and a currently eligible supplier are different things. Follow the numbers behind our October geography tables before comparing them with a vendor list.

Historical snapshot: October 5, 2026. Active and expired registrations are included. This is not a count of all businesses, current eligibility, contracts or spending.

Three counts. Three different questions.

  1. Step 1

    Source rows

    905,712

    Data records in the dated file, before deduplication.

  2. Step 2

    Extra rows removed

    7,508

    Repeated UEIs. Keep the first source row for each UEI.

  3. Step 3

    Distinct UEIs

    898,204

    The denominator for our October geography pages.

905,712 source rows − 7,508 repeated rows = 898,204 distinct UEIs.

These are distinct identifiers, not unique parent companies. For these geography calculations we retain the first row in source order for each UEI, without merging names or locations. The reviewed duplicate rows did not disagree on the geography used here.

Why the lookup has a smaller total

The archived October lookup load kept 898,189 identities outside a reviewed quarantine of 15 incomplete identities. The geography denominator retains all 898,204.

898,189 + 15 = 898,204.

Outside quarantine does not mean every field is complete, or that a registration is currently active. This describes the archived load, not a live database count.

Missing geography stays in the count

The global denominator includes 17 UEIs with no classifiable country: 16 blank fields and 1 malformed code. Fifteen are quarantined; two are outside quarantine.

861,258 USA + 36,929 other valid codes + 17 unclassified = 898,204.

The 203 valid SAM code groups are source labels. They are not a certified count of countries; no country names are inferred.

Reproduce the count without changing the population

  1. Identify the exact file. Use the PUBLIC MONTHLY V2 extract dated October 5, 2026. Match the BOF and EOF control records, date and declared data-row count. Exclude those control records from the data-row total. A different date is a different population.
  2. Read the official layout. This reviewed V2 file has 142 pipe-delimited fields and an !end record terminator. Use the one-based field positions below; do not split a company name or guess a column from its contents.
  3. Deduplicate in source order. Group by UEI and keep the first row. Count retained identifiers and extra rows separately. Do not treat a repeated row as a second supplier or infer corporate ownership from a name.
  4. Separate completeness from geography. Keep the reviewed 15-identity quarantine distinct from the all-source count. Trim and ASCII-uppercase the geographic fields. Accept country groups only for three ASCII letters; retain blank or malformed fields in an unclassified group.
  5. Check the reconciliations above. The twelve published locality labels and eight USA region codes are selected groups, not a complete USA map. Do not sum their shares as if they covered the country. Percentages on the geography pages use their labeled denominators and round to four decimal places.
Field map · one-based positions in the reviewed PUBLIC V2 layout
PositionFieldCounting use
1UEIDeduplication key
18Physical address citySource locality label
19Physical address province or stateSource region code
22Physical address country codeCountry grouping or unclassified
142End of record indicatorRecord-format check (!end)

This page publishes aggregate counts and field definitions. It does not redistribute raw names, identifiers, addresses or contact fields. BROOKLYN and NEW YORK remain source locality labels; no municipal boundaries or spelling variants are inferred.

Use the method in a supplier review

If your list contains 100 suppliers and only 96 match this snapshot, record four unmatched entries for follow-up. Do not label them inactive or excluded. This is an illustrative example, not observed customer data. Check identifier format, identity and source date, then confirm the current official record on SAM.gov.

GSA documents that the monthly public file includes active entities and those expired in the previous six months. Public-display coverage and a dated snapshot limit what a missing match can establish. Entity extracts and exclusions extracts are separate products: these counts do not clear a supplier for exclusions.

Source and audit trail

Source agency: GSA / SAM.gov. Period: October 5, 2026. Geography: the worldwide public-extract population, with selected USA locality views. Unit: source rows or distinct UEIs, as labeled. Documentation consulted October 31, 2026; field layout archived October 7. Counts were reconciled in the October 30 review of the archived file. This page does not claim a newer extract or an automatic update.

GSA entity and exclusions extract documentation · Official PUBLIC V2 field layout · Official SAM.gov entity information

File: SAM_PUBLIC_MONTHLY_V2_20261005.dat

SHA-256: a7beeb182e514f6d17a13ac4e66d1b6223e75c8be42f8cc2c38f155754c19753