A company announces a federal contract "worth up to $480 million." The stock moves. Somebody reads the USAspending record and finds an obligated amount of $2.1 million. Both numbers are real. They measure completely different things, and knowing which one is which is most of what it takes to read a federal award correctly.
Federal procurement data is unusually good by public-data standards. It is free, it has an API, and it covers nearly every contract action the government takes. What it lacks is a plain-language guide to its own vocabulary. This piece is that guide: which fields matter, what each one promises, and where the record goes quiet.
Where the data comes from
When a contracting officer awards, modifies, or ends a contract, the action gets reported into the Federal Procurement Data System. That feed is the raw spine of federal contract reporting. USAspending.gov, run by the Treasury Department, then presents that data with recipient, agency, and account context layered on top. The site exists because Congress required it, first through the Federal Funding Accountability and Transparency Act and later the DATA Act.
Two other official sources sit alongside it. SAM.gov holds entity registrations and most solicitations, so it tells you what the government asked for before anyone won it. The Government Accountability Office publishes bid protest decisions, which are the best free source for understanding why an award was challenged or overturned.
There is no paid product that has data the government does not publish first. Anything commercial is a reshaping of these feeds. Worth remembering before you pay for one.
The two numbers people confuse
Every contract record carries several dollar figures. Three matter most.
Obligated amount is money the government has legally committed. This is the real number, in the sense that it represents a binding commitment against appropriated funds. In the USAspending API it appears as total obligation. It can go up with each modification, and it can go down, which is a detail we will come back to.
Current award amount is the base value plus the options that have actually been exercised. Think of it as the contract as it stands today, including work already triggered.
Potential award amount is the base value plus every option, whether exercised or not. This is the ceiling. It is the number that appears in press releases, because it is the biggest one. The underlying field is base and all options value, and it is a limit on what the government may spend under this vehicle, not a forecast of what it will.
The gap between obligated and potential is often enormous, and that is normal rather than suspicious. A five-year contract with four option years may obligate only the first year at award. Whether years two through five ever get funded depends on appropriations, performance, and whether the agency still wants the work.
Obligations are also not payments. Outlays are payments. USAspending reports outlay data at the award level, but coverage is thinner and lags obligations, so most analysis is built on obligations with that caveat stated openly.
Award types change what the numbers mean
The record will tell you the award type, and this single field reframes everything else.
A definitive contract is a discrete award for defined work. Here the obligated, current, and potential values behave the way intuition expects.
An indefinite delivery vehicle is a different animal. IDIQ contracts, blanket purchase agreements, GWACs, and schedule contracts are all containers. The vehicle itself sets a ceiling and the ground rules. Actual work arrives later as task orders or delivery orders, each with its own obligations, each linked back to the parent award identifier.
This is where the "$480 million contract" usually lives. If the vehicle is a multiple-award IDIQ, several companies hold a seat and compete for individual orders. The ceiling is shared across all of them. A single holder might win a large share of the orders, a token share, or none. The ceiling tells you the size of the room, not the size of anyone's chair.
So the first question on any large award is whether it is a single-award or multiple-award vehicle, and how many awardees exist. The record supports that question directly, and a ceiling read without it is close to meaningless.
The header fields, in order of usefulness
Award ID, or PIID, is the contract's identity. The referenced IDV field points to the parent vehicle when the record is an order under one. Following that link upward is how you see the full structure.
Recipient is identified by a Unique Entity Identifier, which replaced the old DUNS numbers. The parent recipient field maps a subsidiary to its corporate family, which matters because the entity winning the work is frequently a division whose name looks nothing like the listed parent.
Awarding agency and sub-agency say who bought. Funding agency says whose money it is. These differ more often than you would expect, especially for shared service and interagency work.
NAICS and PSC codes describe what was bought, by industry and by product or service. PSC is usually the more precise of the two for services.
Period of performance gives a start date, a current end date, and a potential end date that includes unexercised options. The spread between the last two is the option tail expressed in time rather than dollars.
Type of contract pricing tells you who carries risk. Firm-fixed-price means the contractor absorbed overruns. Cost-plus arrangements mean the government does. Same headline value, very different margin profile.
Modifications, and why totals move
Contracts are amended constantly. Each modification is its own record with a modification number, and the award's totals reflect the accumulated history. Option exercises, scope changes, funding increments, and administrative corrections all arrive as mods.
Negative obligations are real. A deobligation returns committed money that will not be spent, and it shows up as a negative figure. If you sum obligations over a period, you are netting these out, which is usually what you want. If you are counting "new awards," you need to exclude mods or you will count the same contract many times.
A contract that grows steadily through mods and option exercises is telling you something the original announcement could not: the agency kept buying.
Competition fields worth a glance
Extent competed and the number of offers received are quietly among the most informative fields in the record. A competed award with one offer received is a different situation from a competed award with eleven. Sole-source awards carry a stated authority for why competition was not required.
Set-aside type shows whether the award was reserved for small business or a specific socioeconomic category. That constrains who could have bid at all, which changes how you read the competitive picture.
What the record does not show
Subawards are reported separately, by prime contractors, and coverage is uneven. A prime that passes most of the work through to subcontractors will still show the full obligation under its own name.
Profitability is absent. Federal revenue is not federal margin, and a cost-plus contract with a thin fee can look impressive at the top line.
Classified and certain sensitive work is either withheld or reported at reduced detail. Awards can also be reported late or with errors, and corrections arrive as later modifications. The data is good, not perfect.
Timing has one structural quirk worth knowing. The federal fiscal year ends on September 30, and agencies obligate heavily before expiring funds lapse. A September spike in obligations is a budget calendar artifact, not a demand signal.
A short reading procedure
Open the award. Check the award type first. If it is an order, walk up to the parent vehicle and find out how many awardees share it. Read obligated, current, and potential values as three separate facts rather than one number with variations. Look at the modification history to see whether the award has been growing or shrinking. Check the pricing type and the number of offers. Only then decide what the headline figure was actually describing.
Where this sits next to disclosure data
Procurement data is a spending record, not a trading record, and the two answer different questions. Congressional trade disclosures arrive under the STOCK Act with a reporting deadline measured in weeks rather than days, which we cover in detail in the 45-day rule and why it matters. The useful habit is to keep the two separate: contract records tell you what the government committed to buy, and disclosures tell you what an individual reported doing. Neither is a recommendation, and nothing here is investment advice.
If you want the disclosure side of that picture already parsed, scored, and dated, our Congress Stock Trades report turns House and Senate filings into a structured dataset with the filing lag attached to every row, so you can line it up against whatever public spending data you are reading.
Want the signal instead of the raw filings? Get a free report preview. Prefer the tool to the write-up? Browse all data feeds or connect the free MCP server.