Home › Guides › US federal contracting
US federal contracting data: who won what, from which agency, at what value
The US government discloses every federal award by law — the recipient, the amount obligated, the funding agency and the period of performance. That makes federal contracting one of the few markets where you can see your competitors' revenue line by line. This guide covers the two levels of that data, the identifier that makes it joinable, and the traps that produce wrong totals.
Two levels: awards and recipients
USAspending publishes the same spending twice, at different granularities, and choosing the wrong one is the most common mistake:
| Award level | Recipient level | |
|---|---|---|
| One row is | A single contract, grant or task order | A company |
| Key fields | Award ID, title, obligated amount, funding agency, PSC, place of performance, period | UEI, DUNS, address, parent company, business types, lifetime obligations |
| Answers | "What did Palantir win this year, and from whom?" | "How big is Palantir's federal footprint in total?" |
| Dataset | Federal Contract Awards | Federal Contractor Profiles |
A real award row from a measured run:
{
"companyName": "PALANTIR USG INC",
"awardId": "W9128Z26FA001",
"awardTitle": "CDAO MSS TASK ORDER OFF OF CONTRACT W519TC25D0039",
"awardType": "DELIVERY ORDER",
"amount": 545578476.94,
"currency": "USD"
}UEI: the join key that actually works
Matching federal data on company name fails quietly. The legal entity, the brand and the subsidiary holding the contract are often three different strings — PALANTIR TECHNOLOGIES INC. is not what appears on every award. The Unique Entity Identifier replaced DUNS as the key federal systems use, and it is the reliable join across datasets.
In our measured run of recipient profiles, the UEI populated on 93% of rows and the legacy DUNS on the same share — the gap is historical records that predate universal UEI assignment. Where both are missing, the recipient ID still links the row back to its USAspending page.
Four traps that produce wrong numbers
- Double-counting group revenue. Large primes register many entities. Each carries its own totals. Group by parent company before summing, or a group's federal revenue inflates several times over.
- Reading obligation as contract value. The obligated amount on an award record is what has been committed on that record — a multi-year programme is spread across many task orders and modifications. The headline number for a programme is the sum, not any single row.
- Assuming awards carry an award date. They frequently do not: USAspending's award search response does not include one for these rows, so the period-of-performance start is the closest published equivalent. A tool that silently invents a date is guessing.
- Treating grants as contracts. The same search surface covers contracts, grants, loans, direct payments and IDVs. A university's federal income is grants; a defence prime's is contracts. Filter deliberately.
Teaming: finding the partner, not the competitor
Set-aside requirements mean primes constantly need qualified small-business partners. Recipient profiles carry the business-type classifications — small business, veteran-owned, 8(a), women-owned — so filtering on those flags produces a teaming shortlist directly, without reverse-engineering it from award history.
State-level registries complement this: a federal classification tells you what a company is certified as, while a state business registry confirms it exists and is in good standing. See US state business registries for that layer, or check Texas, New York and Connecticut directly.
Adjacent: regulated institutions
One category of federal counterparty has its own richer registry. Banks are supervised rather than merely contracted, so the FDIC publishes their condition — total assets, deposits, equity, net income and the computed ROA/ROE ratios — alongside identity. For anyone selling into financial institutions, that is a far better qualification signal than a contract list. In our measurement the financial fields populated on every active institution; the closure date populated on 68% precisely because the default run deliberately includes failed banks.
Which dataset for which question
| Question | Dataset |
|---|---|
| What did this company win, and from which agency? | Federal Contract Awards |
| How large is their total federal footprint? | Federal Contractor Profiles |
| Which small-business partners are certified in my discipline? | Recipient profiles, filtered on business types |
| Is this bank sound, and how big is it? | FDIC BankFind |
| Does this entity legally exist in its state? | State registries |
| Public contracts outside the US | UK · EU TED · India |
Is this data free to use?
Federal spending records are published under the DATA Act precisely so the public can see where federal money goes. The rows describe organisations and public contracts — business information, not personal data. No API key or registration is required to read them.
Set-asides: who a contract was actually available to
A federal award record carries a field that changes how you read every other number in it: the set-aside type. Many US federal contracts are reserved for specific categories of business — small businesses, and several more narrowly defined designations. A reserved contract was never open to the general market.
This matters in two directions:
- Competitive analysis. A competitor's award total mixes open competition with reserved awards. If a meaningful share came through set-asides they could not lose to you, their apparent strength in open competition is overstated.
- Qualification. If you do not hold a designation, awards reserved for it are not addressable market. Filtering them out gives you a smaller and far more honest picture of the opportunity.
Pair this with the NAICS code on each award, which classifies the work being bought. NAICS is the federal equivalent of the shared taxonomies that make European registers and EU procurement queryable at scale: it lets you define your addressable segment once and apply it consistently across every agency.
Together, NAICS plus set-aside answers the question that raw award totals cannot — not "how much was spent on this", but "how much was spent on this, on work I do, in competitions I could have entered".
Frequently asked questions
Where does the data come from?
USAspending.gov, published by the US Treasury. It is the authoritative disclosure of every federal award, not an aggregator's reconstruction.
Award or recipient — which do I need?
Awards for "what did they win recently"; recipient profiles for "how big are they in total". They answer different questions and mixing them produces wrong totals.
Why does one company appear several times?
Primes register multiple entities, each with its own UEI and totals. Group by parent company before summing.
Can I get grants instead of contracts?
Yes — the same search covers contracts, grants, loans, direct payments and IDVs. Choose the award family deliberately.