Federal IT contract spending data: the records behind a $102B buying ledger

Four records answer the question, and they work at three different grains - which is exactly why teams run more than one.

A fourth record works the edges: Data.gov Catalog - IT Services structured datasets indexes 552,271 datasets across the US federal, state and city estate, and its "information technology services" query surfaces spending, procurement and workforce records as catalog metadata. It is a finding aid, scored 6 - useful for discovery, never a substitute for the three primaries.

A sample row shows the texture:

Two design choices make the corpus unusually well behaved. First, current_uii is a genuine primary key: the same identifier recurs across the portfolio, contracts, projects and evaluation feeds, so one join attaches dollars to vendors to governance grades. Second, the six fiscal-year columns sit side by side on one row - a FY2020-to-FY2025 trend line with no panel assembly.

One caveat belongs in any plan built here. An open letter from Federal CIO Gregory Barbaccia puts the portal into a "streamlined state" from April 2026, refocused on statutorily required data with continued public availability promised. Datadory deliveries hold the corpus as captured, so your historical slices survive whatever the front-end becomes.

Through the contracts report - the bridge that turns a budget line into a counterparty. It maps each investment's current_uii onto contract PIIDs; the Census Bureau's "Field Support Systems" investment (006-000400800), for instance, arrives carrying PIID DOCYA132315BU0042, with reference PIIDs preserving parent and superseded-contract relationships through recompetes.

The entity side matters just as much for prospecting. The registry describes roughly 500,000+ organizations holding Unique Entity IDs, CAGE codes and NAICS/PSC classifications with registration status and business types - which means you can profile who an incumbent is before you chase the contracts they hold. Every join in the chain runs on identifiers: UII to PIID to UEI. Nothing depends on fuzzy company-name matching, which is the failure mode that kills most procurement models at production.

Coverage runs from FY2007 to the current fiscal year, with place-of-performance geography down to county and district - the layer that answers where federal money lands rather than merely who received it. Award-level records key on generated_internal_id, which joins cleanly across transaction, subaward and recipient views, so a bottom-up rebuild reconciles against a top-down rollup instead of arguing with it.

For breadth rather than depth, the Data.gov Catalog - IT Services structured datasets record advertises the wider harvest: 552,271 cataloged datasets with DCAT-US metadata naming publisher, identifier, keywords, modified date and distributions per entry. Treat it as the index card drawer - the way to find an agency dataset you did not know existed, not a place to measure anything.

Five recur enough to name:

The business-development sequence built on this stack is worked through on the sales growth teams page for this industry.

Each record wins a different question, and the scorecard below ranks them the way a desk actually reaches for them.

Sales and growth teams turn bureau codes and contract PIIDs into account plans: who owns the budget, who holds the work, when the work comes up for recompete.

Competitive intelligence and product teams map every federal IT program to its owner and incumbent, then position a displacement case against the government-wide baseline on identical fields.

Journalists, academics and students get statutorily mandated numbers with enough structure to cite: CIO grades beside dollar lines, place-of-performance geography down to county, and a two-decade award trail behind every claim.

Across Datadory's full catalog of 1,744 datasets the mean quality score is 7.81, and this vertical beats it: the 16 primary data processing & outsourced services records average 8.12, anchored by Companies House's perfect 10.

The panel joins cleanly downstream because the keys are declared up front: current_uii inside the investment layer, PIIDs on the contracts report, UEI and CAGE codes on the entity side, NAICS and PSC codes on the rollups - exact matches rather than fuzzy name matching. Source-spelled labels arrive verbatim beside their codes, so nothing normalizes away silently, and extended extracts carry the description, public URLs and mission-delivery classifications that the core roster leaves off.

Start with a sample: name the agencies, investment types or fiscal years you care about, and the extract arrives typed, keyed and joined - the schema you test is the schema every subsequent delivery ships against.

This article is the federal-demand chapter of a wider outsourcing pool. Continue with:

1. data processing outsourced services data guide - the pillar post covering all 22 pooled datasets in the industry, from BPO directories to UK registers to services trade.
2. data-processing-outsourced-services data hub - the full inventory with quality scores, field dictionaries, sample rows and coverage windows.
3. best data-processing-outsourced-services datasets - the top-10 scorecard in one view.
4. Federal IT Dashboard vs IBEF IT & BPM Industry in India - the demand-side and supply-side records head to head.
5. Product pages - start at Federal IT Dashboard for the investment layer, SAM.gov Entity Registration & Contract Awards for transactions, and USAspending.gov - Federal spending by NAICS for rollups.

Then request a sample scoped to the agencies and fiscal years you will actually model - delivered daily, weekly, or hourly, your call.