Federal contracts

USAspending.gov ↗is the U.S. Treasury’s official record of federal spending. It republishes the contract reports that agencies file in the Federal Procurement Data System. We read it through its public API and bulk download. Nothing else feeds this dataset.

We load every award that meets one fixed rule, called r5-v1. An award is in if it passes all three tests:

1. It is a prime contract award: a definitive contract, a purchase order, a delivery or task order, or a blanket-purchase-agreement call. Grants, loans and other assistance are out, and so are the umbrella vehicles that orders are placed under.
2. It was signed in a fiscal year we load. Fiscal years run October 1 to September 30. We load FY2024, FY2025 and FY2026. FY2026 ended on September 30 and is still incomplete because of the reporting delays described below. Each award belongs to exactly one fiscal year, the one in which it was signed.
3. It is big enough. Either the agency reported it as not competed and at least $1 million is obligated, or at least $10 million is obligated however it was awarded.

“Not competed” means the agency coded the award as not competed, not available for competition, not competed under simplified acquisition, or a non-competitive delivery order. That is the same grouping the government uses in its own competition reports, so anyone can reproduce the list on usaspending.gov.

We never choose awards by who received them. Political connections play no part in the selection. Any tag that links a company to a person or a stock is added later by a separate, sourced step, and appears only with its source.

Our award rows hold every large non-competed award, but competed awards only from $10 million up. A share worked out from those rows would overstate the not-competed share. So “percent not competed” always comes from a separate table of agency totals: all prime contract obligations for each agency and fiscal year, and the non-competed part of them, taken from USAspending’s agency figures.

• Every row stores USAspending’s own award ID and a link to that award’s page on usaspending.gov. “View award ↗” opens it.
• A contract number (the PIID) is not unique across the government, so we key rows on USAspending’s award ID, not the PIID.
• Each row records the rule it met and the date of the load that last saw it.

• Obligated means committed so far.

  The dollar figure is the money the government has committed to date on an award signed in that fiscal year. It grows when options are exercised and contracts are modified, so an award signed in FY2024 can show more than a year’s worth of money. It is not the contract’s ceiling value, and it is not money already paid out. Row totals are never labeled “spending in the year.”

• The Defense Department reports about 90 days late.

  DoD’s contract actions show up on USAspending roughly three months after they happen. Our newest fiscal year therefore understates DoD and the government-wide total, and the latest quarter looks smaller than it is. Do not read a drop at the end of the current year as a drop in spending. The data-status page flags this.

• This is a selection, not all spending.

  Non-competed awards under $1 million and competed awards under $10 million are not in the table. Counts and totals of our rows are not government totals.

• Some orders are coded “competed” even though one company got them.

  A task order inherits the competition code of the contract it sits under. Some are placed with one vendor under a “fair opportunity” exception even when that parent contract was competed. The government codes these as competed, so we do too, unless they reach $10 million.

• Amounts move after the fact.

  Awards are modified, de-obligated and corrected. A refresh picks up changes, and each row shows when it was last seen. An award can enter the table later when its obligation crosses a threshold, and leave it if the obligation shrinks.

• Company names are as the government reports them.

  Recipient and parent-company names come from USAspending, not from us. Joining a contractor to a stock ticker is a separate step with its own method, described on the ticker links page.

After each load we compare stored rows with USAspending’s award record on eight fields (amount, recipient, competition code, dates and contract number), and we compare agency totals with USAspending’s own agency figures.

Audit status. In October 2026 the builder checked 30 random FY2026 rows against the award records (all matched) and three agency totals against USAspending’s agency figures (all matched to the cent). A separate review on October 4, 2026 re-checked a sample of awards and the agency totals against USAspending, and the totals matched to the cent. These are samples, so this page publishes no overall match rate. FY2026 totals are not final until the DoD delay has passed.

Found something wrong? Report an error. Past changes are in the corrections log.