What is market research in government contracting?

It has two meanings, and both matter. For the government, market research is a FAR Part 10 obligation: agencies must study the commercial marketplace before writing requirements or soliciting above the simplified acquisition threshold. For a contractor, market research is the work of understanding an agency's buying patterns, incumbents, vehicles, and decision makers well enough to decide whether and how to compete. The two are connected, because the agency's own research generates public artifacts you can read.

When should I start market research on an opportunity?

Well before the solicitation exists. Once an RFP publishes with a 15 to 30 day response window, the requirement is already fixed and the research window has closed. Useful starting points are an expiring contract 12 to 18 months out, a published forecast entry, a sources sought notice, or an advisory committee meeting touching your capability area.

Where can I see which companies subcontract to a prime?

Prime contractors report subawards over $30,000 under the Federal Funding Accountability and Transparency Act. That reporting used to live on FSRS.gov, which GSA retired on March 6, 2025, and now runs through SAM.gov. CLEATUS renders the same filings as prime and subcontractor relationships on contractor and award pages, so you can read the network rather than the raw filings.

Is the Plum Book still published?

Not as a book. Under the PLUM Act of 2022, the quadrennial publication stopped being issued as of January 1, 2026, and OPM's public directory is now the official listing of federal policy and supporting positions. The data covers position title, agency, location, appointment type, and whether a seat is filled, vacant, or held on an acting basis.

Is an AI tool better than manual research for finding government contracts?

It depends on how often you repeat the work. Every source in this guide is free and public, and a careful analyst can answer all five capture questions by hand. We did exactly that for the VA example in this post, and it took an afternoon for one agency and one NAICS code. The manual method is long rather than difficult. What it does not do is scale, because the cost repeats in full for every additional agency, every additional code, and every refresh. Automation changes the arithmetic rather than the answers: the same public records, joined by shared identifiers, so the second agency costs almost nothing after the first. If you research one agency a year, do it by hand. If you are covering a real pipeline, the manual version is where the research quietly stops happening.

How is this different from just searching SAM.gov?

SAM.gov tells you what has been posted. Market research is about what has not been posted yet, and about the context that decides whether a posted opportunity is winnable: who holds the work, how the agency buys, which office signs, and who you would need on the team. Those answers come from award history, subaward filings, meeting records, and leadership data, which sit in separate systems from the opportunity notices.

Do I have to respond to sources sought notices?

You are not required to, but responding is one of the few ways to influence an acquisition before it is written. Your response becomes part of the market research record a contracting officer relies on when deciding between a small business set-aside and full and open competition, and it puts your capability in front of the program office at the point where the decision is still open.

GovCon Market Research in 2026: How to Read an…

Before a federal agency can solicit anything above the simplified acquisition threshold, FAR Part 10 requires it to conduct market research. Someone inside that agency has to survey the commercial marketplace, talk to knowledgeable people in government and industry, review prior acquisitions, and document what they found. That work happens months before the requirement is written and years before an award. It leaves a public trail: sources sought notices, requests for information, industry days, advisory committee minutes, expiring contracts, and subaward filings. GovCon market research is the discipline of reading that trail while the requirement is still being shaped. This guide covers the questions your research has to answer, the public sources that answer them, and why the manual version of this job stopped scaling.

TL;DR

The government does market research before you do. FAR Part 10 obligates agencies to study the market before writing a requirement, and most of that activity is public. GovCon market research means reading the agency's homework, not starting your own from scratch.
A posted solicitation is the end of the research window, not the start. By the time an RFP publishes with a 15 to 30 day response clock, the requirement reflects conversations that happened months earlier.
Five questions decide a pursuit: what is the agency buying, how does it buy, which office signs, who holds the work today, and who would you have to team with. Every one of them is answerable from public data.
The overlooked sources are the useful ones. Advisory committee notices in the Federal Register, local council agendas, subaward filings in SAM.gov, and federal leadership rosters tell you more about a coming procurement than the opportunity feed does.
We ran the framework on one real target to show it works. VA obligated about $2.42 billion under NAICS 541519 in FY2025, 18 of its 25 largest awards in that code are task orders on a NASA vehicle, and all 25 were signed by one contracting office.
Doing this by hand does not scale past a handful of agencies. Four systems, four data models, and no shared identifiers between them. That VA workup took an afternoon, and it covered one agency and one code.
CLEATUS Market Intelligence collapses that workup into one place, from pre solicitation Signals to agency meeting records, leadership charts, contracting office rankings, and subaward networks, with an agent that runs the research and hands back a table, a chart, or a CSV.

Run Your Market Research in One Place

Agency spending, meeting records, leadership charts, contracting offices, and teaming networks, with an AI agent that does the digging for you.

Book a Demo →

The Government Does Market Research First. That Is Your Opening.

FAR Part 10 is short, and it is the most useful five pages in the regulation for a business development team. It requires agencies to conduct market research appropriate to the circumstances before developing new requirements documents and before soliciting offers above the simplified acquisition threshold. It also lists the methods: contacting knowledgeable individuals in government and industry, reviewing recent market research for similar requirements, publishing requests for information, and holding interactive exchanges with industry.

Read that as a schedule. Every one of those methods generates an artifact with a date on it, and most of those artifacts are public. A sources sought notice is the agency asking whether small businesses can do the work. An industry day is the agency describing a requirement it has not written yet. An advisory committee meeting is the agency arguing about the requirement in front of a court reporter.

The contractor who reads those artifacts is doing market research on the government's market research. That is a meaningfully different activity from searching an opportunity feed, and it is the activity that separates a capture plan from a bid decision. FAR 10.001 is explicit that this research shapes what the agency then asks for, including whether commercial products will meet the need and whether the requirement can be set aside for small business. Those decisions are made before anything is posted, and they are made partly out of what industry told the agency.

The Five Questions GovCon Market Research Has to Answer

Market research goes wrong when it becomes a data-gathering exercise with no decision attached. Before you open a single portal, write down which of these five questions you are trying to close.

Question	What answers it	What you do with it
What is this agency actually buying?	Award history by NAICS and product service code, forecasts, meeting records	Decide whether the agency belongs in your pipeline
How does it buy?	Vehicle and set-aside mix, competed versus sole source, average award size	Decide whether you need a vehicle or a partner first
Which office signs the award?	Contracting office rankings, contracting officer notice history	Aim outreach at a buying desk instead of a department
Who holds the work today?	Incumbent award record, period of performance, obligated versus outlaid dollars	Time your capture to the recompete, not the RFP
Who would you have to team with?	Subaward filings, existing prime and sub relationships	Approach a prime with evidence instead of a cold email

If a research task does not move one of those five answers, it is reading, not research. The rest of this guide walks the sources that close them.

A Worked Example: VA Cloud and IT Services, NAICS 541519

Method is easy to describe and easy to skip. Here is the whole framework run against one real target, using only public data. Every figure below comes from USAspending.gov, filtered to the Department of Veterans Affairs as awarding agency, NAICS 541519 (Other Computer Related Services), contract award types A through D, fiscal year 2025, and pulled on August 19, 2026.

What is the agency buying? VA obligated roughly $2.42 billion against NAICS 541519 in FY2025. That is a real market, not a rounding error, and it is the second largest across the federal government in that code behind the Department of Defense.

How does it buy? Through other agencies' vehicles, overwhelmingly. Of the 25 largest VA awards in that filter, 18 are task orders under NASA SEWP V and 5 more sit under GSA vehicles. Only 2 are standalone VA contracts. If you are not on SEWP or a GSA schedule, the honest read is that your first move here is a vehicle or a teaming agreement, not a proposal.

Which office signs? All 25 of those top awards carry a PIID beginning 36C10B, which is VA's Technology Acquisition Center in New Jersey. One office signs effectively all of VA's large IT task orders in this code. That single fact redirects an entire outreach plan.

Who holds the work today? Take one award. Four Points Technology, L.L.C. holds 36C10B22F0207, described as "Enterprise Cloud Capacity - Amazon Web Services," with $521.8 million obligated. Its period of performance runs August 3, 2022 to February 2, 2027, and it was awarded as a service-disabled veteran-owned small business set-aside against 2 offers received. A contract that size ending in early 2027 is a capture target today, not in 2027. It is also a lesson in how the work is packaged: this is a task order for cloud capacity, resold under a set-aside, on a NASA vehicle, signed by a VA office in New Jersey.

Who would you have to team with? The subaward filings on VA's 541519 awards name real relationships. Deloitte & Touche LLP, prime on 36C10B24F0281, reported a $25.8 million subaward to Purple Jay LLC in October 2024. CACI NSS, LLC, prime on 36C10B20F0250, reported subawards to CGI Federal and to Anivas Tech LLC in June 2025. Those are documented doors, and they are a better starting point for a teaming conversation than a cold introduction.

Five questions, one afternoon, one agency, no paid data. What that workup does not tell you is whether the VA program office is already reshaping the requirement, which is exactly what the meeting record and forecast sources below are for.

Source One: The Public Record of an Agency Deciding What to Buy

Closes: what is this agency actually buying, and is the requirement still being shaped.

Most contractors monitor solicitations. Far fewer monitor the meetings where requirements get argued out, and those meetings are published by law.

Federal advisory committees. Under the Federal Advisory Committee Act, an agency has to publish notice of an advisory committee meeting in the Federal Register at least 15 calendar days in advance. The notice states whether the meeting is open or closed under the Government in the Sunshine Act, and it names the Designated Federal Officer you can contact for more information. Agendas, minutes, and transcripts follow. A committee debating a modernization roadmap in March is describing the statement of work you will read in December.

State, local, and education bodies. The equivalent record at the SLED level lives in council and board agendas, often published through Legistar or a similar clerk system. A county commission approving a feasibility study is a procurement in an early stage. If you are moving into state and local work, the SLED expansion playbook covers how those buying cycles differ from federal ones.

Sources sought notices and requests for information. These are FAR Part 10 in its most literal form. A response is not a proposal, and it is not a lot of work, but it puts your capability statement in the market research file the contracting officer will cite when they choose between a set-aside and full and open competition.

Forecasts and expiring contracts. Agencies above a spending threshold are required to forecast upcoming procurements, though quality varies widely. The more reliable signal is arithmetic: every contract has an end date, and a services contract with all options exercised is a successor solicitation waiting to happen. We covered that in depth in the guide to procurement forecasting and recompete intelligence.

Source Two: Award History Tells You How an Agency Buys, Not Just What

Closes: how does it buy, and who holds the work today.

Spend totals are the least interesting thing in award data. The useful patterns are structural, and reading them is what agency-level intelligence is for: the spending trend by award type, the ranked incumbents, and the sub-agency rollups underneath a department.

Competed versus sole source. An agency that awards most of its work in your NAICS through sole source justifications is telling you that relationships and vehicles decide outcomes there, and that a cold proposal will lose. An agency with a high competed share is telling you the opposite.

Vehicle concentration. If most of the dollars flow through one or two IDIQs, your real research question is not "how do I win this contract" but "how do I get on that vehicle, or onto a team that already is." The VA example above is the extreme version, where 18 of the top 25 awards ran through a single NASA vehicle. Reading a vehicle down to its task orders, awardees, and remaining ceiling tells you whether there is room on it.

Obligated versus outlaid. Obligated dollars are what the government committed on paper. Outlaid dollars are what has actually been disbursed. A contract with a large ceiling and slow outlays is behaving differently from one spending down on schedule, and the gap tells you something about how much work is really left.

Award modifications over time. A contract is rarely one transaction. It is a stack of modifications, option exercises, and de-obligations. Reading that history shows you whether the incumbent has been growing scope or losing it, which is why an incumbent should be read as a full federal footprint rather than a single award.

One caveat on tooling: FPDS was decommissioned in February 2026, with contract award data consolidating into SAM.gov. If your research process still assumes an FPDS query, it needs rebuilding.

Source Three: The Subaward Layer Almost Nobody Reads

Closes: who would you have to team with.

Prime contractors are required to report first-tier subawards over $30,000 under the Federal Funding Accountability and Transparency Act. The clause that carries the requirement is FAR 52.204-10, and its full regulatory text sits on acquisition.gov. That reporting moved off the legacy FSRS system when GSA retired FSRS.gov on March 6, 2025, and it now lives in SAM.gov.

This dataset is the closest thing GovCon has to a public teaming graph. It tells you which primes actually hand work out rather than self-performing, which small businesses they hand it to, and roughly how much. For a subcontracting strategy that is far better evidence than a capability matrix on a website. It also works in reverse: if a competitor keeps appearing as a sub under the same three primes, you have learned how they get their work, and whether the same door is open to you.

Read it knowing exactly what it excludes, because the gaps are structural rather than sloppy. Only first-tier subcontracts are reported, so if your target sub works for a sub, that relationship is invisible in this data. A subcontractor whose prior-year gross income was under $300,000 is exempt from the reporting, which removes much of the small end of the market. Supplier and vendor agreements fall outside the definition of a first-tier subcontract, so a company that supplies parts or licenses rather than performing scope will not appear. Add the obvious caveat that filings show dollars and not scope quality, and the honest reading is that a filing proves a relationship while an absent filing proves nothing. Our guide to finding teaming partners in government contracting covers how to turn the relationships you do find into an actual conversation.

Source Four: The Org Chart Behind the Requirement

Closes: which office signs the award, and who inside it you should be talking to.

Knowing an agency buys what you sell is not the same as knowing who to call. Two public sources close that gap.

Federal leadership rosters. The publication known as the Plum Book listed policy and supporting positions across the executive and legislative branches, published every four years after a presidential election. Under the PLUM Act of 2022 that changed: beginning January 1, 2026, the printed publication is no longer issued, and OPM's public directory serves as the official listing. The data is searchable by agency, position title, location, appointment type, pay plan, tenure, and vacancy status, which means you can see not only who holds a seat but which seats are empty and which are filled on an acting basis. A vacant or acting position in a program office is a live variable in your capture plan.

Use it, but do not treat it as complete. GAO reported on February 26, 2026 that the directory covered 10,540 positions across 171 federal entities as of July 2025, while at least seven federal entities and at least 130 presidentially appointed, Senate-confirmed positions were missing from the data entirely. GAO also found errors and inconsistencies within it, including duplicative positions and identifier numbers that were not applied consistently. A missing seat in this directory is not evidence that the seat is vacant, so confirm anything load-bearing against the agency's own site before you build outreach around it.

Contracting officer history. The person who signs the award leaves a trail of notices. What they have solicited before, for which offices, and alongside which colleagues tells you whether you are talking to the right desk. You can browse contracting officers and their notice history on the CLEATUS government people directory.

The Source Table

Here is the whole research surface in one place, so you can see what each system does and does not carry.

Source	What it tells you	Where it lives	Update cadence	Statutory basis
Sources sought and RFIs	That a requirement exists and what the agency thinks it needs	SAM.gov	Continuous	FAR Part 10
Advisory committee notices	What the agency is debating, and who is in the room	Federal Register	At least 15 days before each meeting	FACA, 41 CFR 102-3.150
Local meeting agendas	The same signal for state, local, and education buyers	Legistar and similar clerk systems	Per meeting, set by local law	State open-meetings statutes
Procurement forecasts	Planned buys the agency chose to publish	Agency forecast pages, GSA	Annual, quality varies widely	Small Business Act 8(a)(12)(C)
Prime award history	Incumbent, dollars, vehicle, set-aside, period of performance	USAspending.gov and SAM.gov	Daily	FFATA
First-tier subaward filings	Which primes hand out work, to whom, for how much	SAM.gov, formerly FSRS.gov	Monthly, by the reporting prime	FAR 52.204-10
Federal leadership roster	Who holds a seat, and which seats are vacant or acting	OPM public directory	Annual agency reporting	PLUM Act of 2022

Manual Market Research Versus an Automated Workup

None of the sources above are secret. The problem is that they live in different systems, with different identifiers, different update cadences, and no join key between them. Searching SAM.gov alone is not enough, because SAM.gov carries what has been posted, and four of the five questions are answered somewhere else. We break down that gap in detail on our SAM.gov comparison page.

A realistic manual workup on a single target agency looks like this: pull award history and filter it by NAICS, then pull subaward filings separately and try to match company names that are spelled four different ways, then search the Federal Register for the relevant advisory committees, then find the agency's own forecast page, then look up the contracting office in the award records, then search for the program office leadership by name. Nothing links to anything.

Here is the same five-question workup run both ways. The manual column is what we actually did to produce the VA example above.

Research step	By hand	With an agent	What changes
What the agency buys	Filter award data by agency, NAICS, award type, and fiscal year, then total the obligations	Ask for the spend in your code and get the figure with its trend	Very little. This step is quick either way.
How it buys	Open the largest awards one at a time and read each parent vehicle off the record	Vehicle mix returned as a ranked breakdown, with the remaining room on each	The SEWP concentration is only visible after you have opened 25 records. It is the finding that changes the plan.
Which office signs	Read the PIID prefix off each award and map it to an office yourself	Contracting offices ranked by award dollars inside the agency	`36C10B` is obvious once ranked and invisible while you read PIIDs one by one.
Who holds the work	Record the incumbent, obligation, period of performance, set-aside, and offers received per award	Incumbents with end dates, plus recompetes predicted from award history	By hand you find the contract. The end date that makes it a target still has to be tracked somewhere.
Who to team with	Query subaward filings per prime award, then reconcile company names spelled several ways	The first-tier network walked in both directions, from the prime or the sub	Name reconciliation is most of the cost here, and it is the step people abandon first.
One agency, one code	An afternoon	Minutes, and the next agency needs no new setup	Manual cost is linear per agency and per refresh. Automated cost is close to flat.

Read the last row carefully, because it is the whole argument. No single step above is hard. The manual workup is long rather than difficult, and its cost repeats: every additional agency, every additional NAICS code, and every quarterly refresh costs the same afternoon again. A team that also has proposals to write runs out of afternoons before it runs out of agencies worth researching.

That is why many small and mid-sized contractors skip the research and go straight to bidding, which is also why so many proposals arrive with no idea who the incumbent is. It is not a discipline problem. It is a tooling problem, and it is the same one that made capture management look completely different before AI.

What the market uses today

Several tools already do parts of this job, and it is worth knowing what each one is for. GovWin IQ is the long-established enterprise option, strongest on human-collected pre-RFP intelligence and state and local coverage, priced and scoped for larger capture organizations. HigherGov is the widely used research and data platform, deep on award history, contractor profiles, and vehicles, and the tool many small businesses reach for first (how it compares to CLEATUS). GovTribe covers similar ground from the agency side, tracking awards, vehicles, and the people attached to them, with its AI features priced as usage-based credits (GovTribe compared). Govly answers a different part of the question: it is a teaming and private contract vehicle network, which matters once your research has told you which partner you need rather than which agency to study (Govly compared). The practical difference is where each sits in the timeline: research platforms tell you what has already happened, opportunity feeds like SAM.gov tell you what has posted, and the harder problem is connecting both to what an agency is about to do.

How CLEATUS Runs This Research for You

The VA workup earlier in this post is a job, not a search. It required a filtered award pull, a grouping by office, a read of one award's terms, and a separate subaward query, and each step depended on the answer before it. That shape is what the GovCon Copilot is built to run: you state the objective, and it plans the steps, calls the tools it needs across agencies, contractors, contracting officers, and award history, and keeps working until the question is closed. It is not a chat box that summarizes a page you already found.

So the five-question workup becomes something you assign rather than perform:

"Run a market research workup on the VA for NAICS 541519: spend, vehicle mix, top contracting offices, largest incumbents with their end dates, and who they subcontract to."
"For this agency, what share of awards in my NAICS were sole source versus competed?"
"Show me the primes that subcontract most often in this NAICS, ranked by disclosed subaward dollars."
"Which contracts in my space expire in the next 18 months, and who holds them?"
"Summarize this agency's advisory committee activity over the last year and flag anything touching my capability."

The answer comes back as a table, a chart, or an export you can hand to someone, not a wall of links. If you keep prompts like these around, the Prompt Library turns them into a shared team asset instead of something one person retypes.

What makes that possible is the evidence layer underneath, where the four public sources in this guide already sit on the same pages, connected by the same identifiers. CLEATUS Market Intelligence carries Signals for work that has not posted yet, covering subcontract scopes inside a prime award, agency forecasts, and recompetes predicted from award history, each on a page built for what it actually is. Agency Meeting Signals puts the Federal Register and Legistar meeting record on the agency page, including cancelled sessions, since a scrapped meeting is also information. Agency Org Charts map leadership down to individual seats with filled, vacant, and acting tallies, and Top Contracting Offices ranks the desks that sign, which is how you find a 36C10B without reading 25 PIIDs by hand. Subcontractor and teaming relationships render the first-tier filings as a network you can walk in both directions.

The agent reads all of that. The pages are there for when you want to check its work, which you should.

"CLEATUS fundamentally changed the way we capture, analyze, and build proposals. We tripled our output without adding staff, and the platform finally moves at the speed our workflow demands."

– John Garnish, Business Development Lead, D2 Government Solutions

D2 reports a 75% reduction in opportunity discovery time and an 80% reduction in draft time. MST Maritime Management went from 3 proposals a month to more than 10. In both cases the gain came from the front of the process, where research either narrows the field or does not.

One practical note: the research surface described above sits in the CLEATUS DATA plan, published at $39 per user per month, $468 per year. That covers Market Intelligence, agency and contracting officer data, contractor and award history, forecasts and pre-forecasts, and unlimited contract and government data search.

Frequently Asked Questions

Do the Research Before the Clock Starts

The contractors who win consistently are not the ones who read solicitations faster. They are the ones who arrive at the solicitation already knowing the agency, the incumbent, the buying office, and the team they would need, because they read the public record while the requirement was still being written. That record is open to everyone. The only real barrier has been how long it takes to assemble.

Book a Demo → to see agency spending, meeting records, leadership charts, contracting offices, and teaming networks on one page, with an agent that runs the research for you.

Customer Stories

About CLEATUS

CLEATUS is an agentic AI platform that helps government contractors discover the right opportunities, manage capture pipelines, and write winning proposals. Automate your GovCon operations end-to-end: build custom AI-powered automations that handle multi-step processes on autopilot, a force multiplier for your capture and BD team that works in the background while you stay in the loop on every decision. We aggregate federal, state, local, and city opportunities; our GovCon Copilot analyzes solicitations and your internal documents to deliver actionable market intelligence that drives revenue growth.

GovCon Market Research in 2026: How to Read an Agency Before the Solicitation Exists