A scheduled pre-solicitation feed is a saved-filter pull against the federal contract opportunities API, run on a fixed cadence, persisted run over run so you can diff what changed, and routed to a named owner for each notice type. Sources sought notices and RFIs become shaping and teaming triggers under FAR 15.201 [3]. Presolicitation notices and synopses published under FAR Subpart 5.2 become schedule and go/no-go triggers [2]. The feed replaces portal refreshing with a process that has owners, cutoffs, and a record of who saw what.
Here is the part most teams miss. The agency has already done its homework by the time a solicitation posts. FAR Part 10 directs agencies to conduct market research to shape requirements and assess commercial availability before the requirement is finalized [1]. That research is the window where your capability language can influence scope, set-aside structure, and evaluation emphasis. Once the RFP is on the street, you are responding to decisions somebody else helped make.
I have watched two bidders see the same sources sought notice on the same day. One logged a shaping response, pulled the incumbent's award history, and knew the period of performance pattern before the notice closed. The other filed the notice in a shared inbox folder. Ninety days later they wrote very different proposals, and only one of them read like it had been in the room.
The RFP Was Never the Starting Line
The regulatory structure tells you where the use point sits, if you read it in order. FAR Part 10 establishes that market research precedes the acquisition strategy, not the other way around [1]. FAR Subpart 5.2 governs how proposed contract actions are publicized, which means the notice stream is a regulated artifact with predictable timing rather than informal chatter [2]. FAR 15.201 then explicitly encourages exchanges with industry before receipt of proposals [3].
That third citation matters operationally. When a capture lead asks whether it is acceptable to engage an agency before a solicitation exists, FAR 15.201 is the answer you point to [3]. The exchange is contemplated by the rules. What is not contemplated is treating the exchange as a sales call.
Sources sought responses also do regulatory work beyond your own pipeline. Agencies use market research under FAR Part 10 together with small business set-aside analysis, including the total small business set-aside rule at FAR 19.502-2, to decide how a requirement will be competed [1][7]. Your response either supplies the agency with evidence of capable small business interest or it does not. Vague capability language gives a contracting officer nothing usable.
So the practical reframe is this: the shaping window opens the moment a notice posts, and the clock on it runs faster than any proposal clock. The notice may close in ten business days. A solicitation gives you thirty. The upstream window is shorter and worth more, which is exactly why it cannot depend on somebody remembering to check a portal.
Why Inbox Monitoring Keeps Losing Notices
Manual monitoring fails in three predictable ways, and all three are invisible until they cost you a bid.
The first is missed amendments. You see the original notice, you make a decision, and then the response date moves. Nobody catches it because the email alert already fired once. The second is duplicated triage work: three people independently open the same notice, form three private opinions, and none of it gets written down. The third is the absence of a record. When leadership asks why the team did not respond to a notice, there is no timestamp showing when it arrived, who triaged it, or what the hold reason was.
Email saved searches tell you a notice exists. They do not tell you what changed since the last version, which is the information that actually drives action.
| Monitoring method | What it catches | What it misses | Recommendation |
|---|---|---|---|
| Manual portal refresh | Whatever happens to be on screen that day | Amendments, cancellations, off-cycle postings, any audit trail | Avoid as a primary method |
| Email saved searches | First-time notices matching broad filters | Field-level changes, revised response dates, duplicate suppression | Acceptable backup only |
| Third-party aggregator | Broad coverage with a built-in interface | Filter transparency, raw response retention, your own diff logic | Useful for discovery, not for pipeline control |
| Scheduled API retrieval | New, amended, and cancelled notices with persisted raw responses | Nothing, if you persist each run and diff on the notice identifier | Primary method for any notice already in the pipeline |
Notice type should drive routing, not just alert volume. Sources sought and RFIs go to the capture lead because they are shaping and teaming decisions under FAR 15.201 [3]. Presolicitation notices and synopses published under FAR Subpart 5.2 go to the pipeline owner because they set schedule and force a bid decision [2].
Without persisted raw responses, you cannot prove a response date moved. That is how teams miss cutoffs and then argue about whether the date was ever different.
Filters, Cadence, and the Files You Keep
GSA publishes a documented public API for contract opportunities, which is what turns retrieval into an operational process instead of a browsing habit [4]. Read the documentation before you build anything, because the parameter structure shapes how you design saved filters.
Build the filter set around five dimensions and version it:
- NAICS codes you can actually win under, not every code in your SAM profile
- Product and service codes (PSC) for the work type, which often catch notices NAICS misses
- Agency and office, because office-level patterns are where buying behavior actually lives
- Set-aside type, matched to your verified size status rather than aspiration
- Place of performance, when staffing geography is a real constraint
Version the filter set in the same place you keep the pull configuration, so a change in results six weeks from now is explainable. A filter nobody can reconstruct is a filter nobody trusts.
filter_set: hhs-health-it-v4
naics: [541512, 541511, 541519]
psc: [D302, D307, D399]
agency_office: [HHS/CDC, HHS/NIH-OALM]
set_aside: [SBA, SDVOSBC, NONE]
place_of_performance: [GA, MD, DC, REMOTE]
notice_types: [sources_sought, presolicitation, combined_synopsis]
cadence: daily
owner: capture_lead_healthPersistence is the part teams skip and later regret. Store the raw API response for every run, keyed by run timestamp. That gives you a diffable history: new notices, amended notices, cancelled notices, and reposted notices under a new title.
Cadence should match intent. Run active filters daily. Run exploratory filters weekly, since their job is discovery rather than schedule control. Run a separate re-pull against every notice already in the pipeline, regardless of whether it still matches your filters, because a notice whose NAICS changed will silently drop out of a filtered pull.
Dedupe on the notice identifier. When a requirement is reposted under a new title, identifier-based matching keeps it as one pursuit instead of spawning a second pipeline record with its own half-finished triage.
Everything above describes the federal notice stream. If your pipeline also covers state, local, and education buyers, our pre-RFP guide for SLED vendors covers monitoring those markets, where opportunity publication is spread across many separate portals rather than one federal API.
Alert on Amendments, Not First Sightings
The highest-value alert in the entire feed is not a new notice. It is a change to a notice you already decided to pursue.
Diff at the field level and alert on six things: response date, set-aside designation, NAICS or PSC change, attachment added or replaced, point of contact change, and cancellation. Each one carries a different signal and a different owner.
A revised response date is a schedule change that invalidates your internal cutoff. A new set-aside designation can change your position from prime to subcontractor overnight. A new or replaced attachment often contains the draft performance work statement, which is the single most useful document available before a solicitation exists. A point of contact change tells you the requirement moved desks, which usually means the timeline moved too. A cancellation ends your spend. A repost with a new identifier needs manual linkage so your shaping history follows the requirement.
Read the direction a date moves, not just the fact that it moved
A response date that moves earlier is a pure schedule risk: your cutoff just compressed and somebody needs to know today. A date that moves later usually means the agency did not get the market response it needed. That is a shaping opportunity, not an inconvenience. Thin response volume strengthens the case that the market needs a different structure, and your capability statement is now one of a smaller set the contracting officer is reading for the FAR Part 10 record [1].
Route amendment alerts to two named people: the capture lead who owns the shaping decision and the pipeline owner who owns schedule. Do not route them to a shared inbox. Shared inboxes are where accountability goes to die, and an amendment alert nobody owns is functionally the same as no alert.
Set a 24-hour action standard on amendment alerts. Not a response standard, an acknowledgment standard. Someone confirms they saw it and records what, if anything, changes in the pursuit.
Enrich Before You Spend
Every notice that survives triage should be joined to award history before anyone drafts a word. The USAspending API and FPDS both expose historical award data programmatically [5][6]. What you are looking for is specific: incumbent name, contract vehicle, period of performance pattern, and prior competition type.
That enrichment changes bid decisions. A requirement that has been single-award and sole-source twice running is a different pursuit from one that has rotated between three small businesses. Our write-up on finding your edge before the RFP drops covers how to turn those patterns into positioning.
Then run the eligibility screens, because shaping money spent on an ineligible pursuit is simply gone.
Triage screens with pass or hold gates:
- Entity registration. FAR Subpart 4.11 sets registration conditions tied to award eligibility [8]. Registration status, expiration, UEI, legal name consistency, and exclusion status can all be verified programmatically through GSA's entity management and entity extract APIs [9][10]. Hold the notice until the check passes. Primes and evaluators can query the same records you can, so a stale record is externally visible.
- Representations and codes. Match the NAICS and size representations on file against the code you are actually pursuing. A mismatch found at submission is a self-inflicted wound.
- Size and set-aside status. Determine status under SBA size regulations at 13 CFR Part 121, including affiliation effects of your teaming structure, which teams routinely underestimate when partners are added [16]. Then read the agency's likely set-aside analysis under FAR Part 19 and the total small business set-aside rule at FAR 19.502-2 so you know whether to pursue as prime or sub [17][7].
- Cyber readiness flag. Scan notice text and attachments for references to controlled unclassified information, safeguarding, assessment, or certification. Those map to the CMMC program framework at 32 CFR Part 170, the CUI requirements in NIST SP 800-171 Rev. 3, and the DoD safeguarding and incident reporting clauses in the DFARS at 48 CFR Chapter 2 [11][12][13]. Route the flag to a technical owner, not the proposal manager.
If a scoped cyber gap cannot be closed before the anticipated solicitation date, that is a prime-versus-sub decision made now, not a discovery made in proposal week. Our CMMC proposal compliance workflow walks through the evidence mapping that supports it.
Feed the screen results straight into your go/no-go scoring framework so the gate is a recorded decision rather than a hallway conversation. Projectory's go/no-go analysis for RFP bid decisions keeps that scoring and its evidence in one place, so a pre-solicitation hold reason is still readable when the solicitation finally posts.
One Pipeline Record, Keyed to the Notice ID
Every shaping touch gets logged against the notice identifier. That identifier is the spine that connects a sources sought response in March to a proposal in August to an award in November.
Minimum fields per record:
- Notice ID and notice type
- Agency and office, at office-level granularity
- Stated response date and your internal cutoff
- Owner name, not team name
- Eligibility screen result with run date
- Identified incumbent and source of that finding
- Response submitted date and the artifact submitted
- Agency contact and channel used
For the response itself, structure beats polish. Answer the agency's stated market research questions in the order asked, because the contracting officer is building a FAR Part 10 record and reads for those answers specifically [1]. State your size status accurately. Cite relevant scope performed with contract numbers, vehicles held, and actual capacity in staff and geography. Keep every claim evidenceable on request.
What gets a response set aside is generic marketing language. A paragraph about commitment to mission does nothing for a set-aside determination under FAR 19.502-2 [7]. Three sentences naming a comparable contract, its value range, and the staffing model does.
This is where a maintained content reuse library pays for itself. Capability statement blocks, past performance summaries, and vehicle lists should be assembled from approved source content rather than rewritten under deadline. The same discipline carries into proposal phase through compliance matrix automation, where requirement-to-response traceability starts from structured source material.
One record, one owner, one identifier. Everything else is reconstruction.
Four Metrics That Tell You the Filter Is Wrong
Instrument the feed or you will never know whether it works. Track four things:
- Notices ingested per filter set. Measures reach. Rising volume with flat responses means the filter is too broad.
- Responses submitted. Measures conversion from triage to action. If this is near zero, either the filter is wrong or nobody owns the triage step.
- Notices that later converted to a solicitation. Measures filter precision against real acquisitions.
- Wins traceable to a logged pre-solicitation touch. Measures whether shaping actually influences outcomes.
Review the first two monthly and the second two quarterly, since conversion lags by a full acquisition cycle. The decision each metric drives is concrete: ingestion volume tunes filter breadth, response rate tunes staffing, conversion rate tunes which NAICS and offices stay in the active set, and traceable wins justify the shaping budget.
Filter precision beats filter volume every time. A filter that returns 200 notices and produces zero responses is noise with a schedule attached. Cut it or narrow it.
Close the loop with post-award data. FAR Subpart 15.5 frames debriefings in negotiated acquisitions, and GAO protest practice runs under its regulations at 4 CFR Part 21 [14][15]. Both produce structured findings. Recurring weaknesses, whether in past performance depth or technical approach specificity, should route back into next quarter's shaping responses and capability statement blocks. Our debriefing runbook covers how to capture those findings as queryable fields instead of meeting notes.
Frequently Asked Questions
What is a sources sought notice, and how is it different from an RFI or a presolicitation notice?
A sources sought notice is a market research instrument. The agency is asking whether capable sources exist, often to support set-aside analysis under FAR Part 19 and FAR 19.502-2 [17][7]. An RFI gathers information on approach, pricing structure, or technical feasibility. A presolicitation notice or synopsis published under FAR Subpart 5.2 announces an intended contract action and signals schedule [2]. The first two are shaping triggers. The third is a go/no-go trigger.
Can responding to a sources sought notice disqualify us or create a conflict of interest?
Responding to market research notices is contemplated by the rules. FAR 15.201 encourages exchanges with industry before receipt of proposals [3]. Risk comes from the content, not the act: inaccurate size representations, claims you cannot evidence, or participation in drafting requirements you later bid. Have counsel review any engagement that goes beyond answering the agency's stated questions.
How often should the feed run, and what cadence is overkill?
Daily for active filters, weekly for exploratory filters, and a separate daily re-pull for anything already in the pipeline. Hourly polling adds cost without adding decisions, since notice postings and amendments do not move faster than a one-day action standard.
What do we have to verify before spending money on a response?
Entity registration comes first. FAR Subpart 4.11 sets the registration conditions tied to award eligibility [8], and GSA's entity management and entity extract APIs let you check registration status, expiration, UEI, legal name consistency, and exclusion status programmatically rather than by eyeballing a profile [9][10]. Then confirm the NAICS and size representations on file match the code you are pursuing, and determine size status under SBA's size regulations at 13 CFR Part 121, including the affiliation effects of your teaming structure [16].
How do we identify the incumbent from a sources sought notice?
The notice itself rarely names one, so join it to historical award data instead. The USAspending API and FPDS both expose prior award records programmatically [5][6], which gives you incumbent name, contract vehicle, period of performance pattern, and prior competition type. Run that join before capture budget is committed, because a requirement that has been sole-source twice running is a different pursuit from one that has rotated among several small businesses.
Do we have to respond to every notice our filters return?
No, and trying to is how shaping budget disappears. Set a response rate target per filter set and treat a low rate as a filter problem rather than a discipline problem. Every notice still needs a recorded triage outcome, including hold reasons, so the decision is visible later.
What is the single next step?
One filter, one owner, one logged response record keyed to the notice identifier.
Start With One Filter This Week
Go back to the two bidders who saw the same notice. The difference between them was not budget or relationships. It was that one of them had a process that produced a record.
Build that process at the smallest possible scale this week. Pick your strongest NAICS code. Build one saved filter against the GSA contract opportunities API with agency, PSC, and set-aside parameters set [4]. Schedule one daily pull and persist each raw response. Name one owner for triage and one owner for amendment alerts.
Then track one metric for a quarter: the percentage of notices you responded to that later converted into a solicitation. That number tells you whether your filter is pointed at real acquisitions or at noise. Everything else in the feed, the enrichment joins, the eligibility gates, the response templates, gets built on top of a filter you already know works.
References
- [1]FAR Part 10 - Market Research (Acquisition.gov). https://www.acquisition.gov/far/part-10
- [2]FAR Subpart 5.2 - Synopses of Proposed Contract Actions (Acquisition.gov). https://www.acquisition.gov/far/subpart-5.2
- [3]FAR 15.201 - Exchanges with Industry Before Receipt of Proposals (Acquisition.gov). https://www.acquisition.gov/far/15.201
- [4]GSA Get Opportunities Public API Documentation (open.gsa.gov). https://open.gsa.gov/api/get-opportunities-public-api/
- [5]USAspending API Documentation. https://api.usaspending.gov/
- [6]Federal Procurement Data System (FPDS). https://www.fpds.gov/fpdsng_cms/index.php/en/
- [7]FAR 19.502-2 - Total Small Business Set-Asides (Acquisition.gov). https://www.acquisition.gov/far/19.502-2
- [8]FAR Subpart 4.11 - System for Award Management (Acquisition.gov). https://www.acquisition.gov/far/subpart-4.11
- [9]GSA SAM Entity Management API Documentation (open.gsa.gov). https://open.gsa.gov/api/entity-api/
- [10]GSA SAM Entity Extracts API Documentation (open.gsa.gov). https://open.gsa.gov/api/sam-entity-extracts-api/
- [11]32 CFR Part 170 - Cybersecurity Maturity Model Certification Program (eCFR). https://www.ecfr.gov/current/title-32/subtitle-A/chapter-I/subchapter-M/part-170
- [12]NIST SP 800-171 Rev. 3 - Protecting Controlled Unclassified Information in Nonfederal Systems (NIST CSRC). https://csrc.nist.gov/pubs/sp/800/171/r3/final
- [13]DFARS - 48 CFR Chapter 2, Defense Acquisition Regulations System (eCFR). https://www.ecfr.gov/current/title-48/chapter-2
- [14]FAR Subpart 15.5 - Preaward, Award, and Postaward Notifications, Protests, and Mistakes (Acquisition.gov). https://www.acquisition.gov/far/subpart-15.5
- [15]4 CFR Part 21 - GAO Bid Protest Regulations (eCFR). https://www.ecfr.gov/current/title-4/chapter-I/part-21
- [16]13 CFR Part 121 - SBA Small Business Size Regulations (eCFR). https://www.ecfr.gov/current/title-13/chapter-I/part-121
- [17]FAR Part 19 - Small Business Programs (Acquisition.gov). https://www.acquisition.gov/far/part-19