Industry Solutions|October 1, 2026|12 min read

Instant Recompete Readiness: Turn Incumbent Performance Data Into Proposal Assets

Incumbents lose recompetes because their best evidence sits in disconnected systems. Here is how to govern CPARS, deliverable, and obligation data as a continuously proposal-ready dataset.

James Whitfield|Federal Capture Manager

It is 30 days to the recompete response date and your team is still reconstructing deliverable dates from email threads and three-year-old slide decks. The program manager who ran the first two option years left in March. Nobody can say with confidence what the actual on-time delivery rate was in FY24, so the past performance volume gets a number that sounds right and a hedge that sounds defensive.

Recompete readiness means governing your incumbent performance data as a live dataset: named owners, a refresh cadence, and evaluator-facing attributes on every record. The attributes that matter are the ones FAR 15.305 tells evaluators to use when judging relevance, namely scope, magnitude of effort, and complexity [2]. The evidence itself gets produced during performance and recorded in CPARS under FAR Subpart 42.15, where it stays available for use in future source selections [1].

Here is the uncomfortable part. The incumbent advantage most teams describe as "relationship" is actually evidence density, and evidence density is a data problem. You either hold your own performance record in retrievable, attribute-tagged form or you rediscover it every cycle at the worst possible moment.

The Recompete You Already Won and Then Lost

Three weeks of reconstruction is not a staffing problem. It is a symptom of treating past performance as a writing exercise that begins when the solicitation posts.

FAR Subpart 42.15 requires agencies to prepare and record contractor performance evaluations in CPARS, and it makes that information available for use in future source selections [1]. Read that again from the evaluator's seat. The narrative an agency wrote about your performance in option year two is standing source-selection evidence, whether or not you ever read it. Your proposal narrative competes with that record; it does not replace it.

FAR 15.305 then directs evaluators to assess past performance relevance in terms of scope, magnitude of effort, and complexity, along with recency [2]. Those are not stylistic preferences. They are filter criteria. When an evaluation team has six references to triage and a scoring sheet to complete, a reference that clearly states its dollar magnitude, period of performance, and technical scope gets credited. A reference that buries the same facts in three paragraphs of adjectives gets a lower relevance finding, not because the work was weaker but because the relevance was harder to establish.

The teams that lose recompetes rarely lose on capability. They lose on traceability. Someone asks "where did the 97 percent figure come from" during Red Team, nobody can answer within the review session, and the number gets softened to "consistently high" before submission. That edit costs you a discriminator.

Where Your Strongest Evidence Actually Lives

Before you build anything, inventory the sources. There are usually five, and they almost never share an owner.

  • CPARS evaluation records, which contain the government's own assessment language by factor [1][4]
  • Contract deliverable and modification history, which establishes what was actually required and when it changed
  • Obligation data, published in machine-readable form through the USAspending API [6]
  • Internal program reporting, including service-level metrics, incident history, and delivery dashboards
  • Staffing history, which supports key personnel continuity and transition claims

Each of these needs a named owner and a refresh cadence assigned before any platform work starts. Unowned feeds decay quietly. A quarterly pull that nobody checks produces stale data that looks current, which is worse than no data at all because writers trust it.

SourceSystem of RecordOwnerRefresh CadenceProposal Use
CPARS evaluationsCPARS [4]Contract managerMonitor continuously for new postings [1]Factor-level rating evidence, government-authored language
Contract modsContract fileContracting leadOn every modificationScope growth, option exercise history, complexity markers
Obligation dataUSAspending API [6]Capture analystMonthly scheduled pullDollar magnitude, obligation trend, period of performance
Delivery metricsProgram reporting systemProgram managerMonthlyQuantified outcome statements with a traceable basis
Staffing rostersHR or PMO systemProgram managerQuarterlyKey personnel continuity, retention, transition claims
Award and notice dataSAM.gov data services [5][9]Capture analystScheduled retrievalVehicle, agency, competitive context

Notice what is absent from that table: slide decks. A deck is a presentation artifact, not a system of record. It has no refresh date, no owner after the meeting ends, and no way to tell you whether the number inside was ever validated. When a portfolio runs eight or ten contracts, reconstruct-every-cycle behavior compounds into a permanent tax on the proposal team.

The failure symptom is easy to spot. Three versions of the same outcome number appear in three volumes, none of them traceable to a system of record, and the Gold Team has to pick one by vote.

Model the Record the Evaluator Is Actually Scoring

Turn the FAR 15.305 relevance language into literal data fields [2]. Not a template with prompts. Fields, with types and validation.

A recompete-ready past performance record should carry, at minimum:

  • Contract or order number and unique entity identifier as join keys
  • Agency and sub-agency, plus contract vehicle
  • Period of performance, start and end, with option years broken out
  • Dollar magnitude, with obligated value separate from ceiling value and the source noted [6]
  • Scope descriptors as a controlled vocabulary, not free text
  • Complexity markers such as number of sites, security environment, integration count, or concurrent task orders
  • CPARS rating by factor, with the evaluation date [1][4]
  • Quantified outcome statements, each tagged with source system, metric definition, and date
  • Named owner and last refreshed date

Here is a sample record field list, the way it should look in a governed library:

record_id: PP-2024-0117
contract_number: 47QTCA21D00XX / Order 0004
uei: ABC123DEF456
agency: GSA FAS
vehicle: Multiple Award Schedule
pop_start: 2021-09-27
pop_end: 2026-09-26
obligated_value: $14.2M   [source: USAspending API, pulled 2026-09-05]
scope_tags: [network_operations, tier2_service_desk, ATO_sustainment]
complexity: 6 sites, CUI environment, 3 concurrent task orders
cpars_factors: Quality=Very Good; Schedule=Satisfactory;
               Mgmt=Very Good; eval_date=2026-03-14
outcome_1: "Tier 2 resolution within SLA, FY25"
          [source: ServiceNow monthly report, validated 2026-08-31,
           approver: K. Ortega, PM]
owner: K. Ortega
last_refreshed: 2026-09-05

The point of the structure is that relevance matching becomes a query. Give me every record where agency equals DHS, obligated value is above $10M, scope tags include ATO sustainment, and period of performance ended within three years. That runs in seconds. The judgment call that used to consume the first four days of a ten-day turn turns into a filtered list a capture manager reviews over coffee.

Narrative-only entries cannot be filtered. That is the whole objection. If a record exists only as prose, every writer rewrites it to fit the new solicitation and every reviewer re-approves it from scratch, cycle after cycle. Attribute tagging is what converts a one-time writing effort into a reusable asset, which is the same principle behind any functioning content reuse library for proposal teams.

Your Competitors Can Already Read Your Contract Record

Award, modification, and obligation data for your incumbent contract is public and machine-readable. GSA publishes documented SAM.gov data services [5], and USAspending exposes award and obligation records through its API [6].

So does your competitor's capture analyst. The advantage does not belong to whoever generated the data. It belongs to whoever ingests and governs it first.

Use the contract or order number together with the unique entity identifier as stable join keys between public award data and your internal delivery metrics. That join is what produces a defensible quantified outcome statement. "We sustained operations across six sites" is a claim. "We sustained operations across six sites under $14.2 million in obligated value across four option years, with monthly service-level reporting from the program's incident system" is evidence with two traceable sources, one public and one internal.

Run the join on a schedule, not on demand. Monthly works for obligation data. The output should land in the same governed library your writers already use, with the source and pull date attached to each value. When a reviewer challenges a number during a color team pass, the answer is a field lookup rather than a research task.

Partition the library before you open it

FAR 52.204-21 establishes basic safeguarding requirements for covered contractor information systems [3]. Performance data pulled from program systems frequently contains proprietary material, and anything touching source selection needs separate handling. Decide access tiers before writers get accounts, not after someone pastes protected content into a reusable boilerplate section. Retrofitting access control into a shared library is far more expensive than designing it in.

The CPARS Window Most Teams Sleep Through

FAR 42.1503 provides a defined period for the contractor to review and comment on a performance evaluation before it becomes final [1]. That right is worth exactly as much as your monitoring discipline.

An unanswered marginal rating does not stay a local disappointment. Under FAR Subpart 42.15, recorded evaluations remain available for use in future source selections [1], which means a factor narrative you never responded to becomes standing evidence against you in competitions you have not identified yet. Teams discover this during a recompete, reading a three-year-old CPARS narrative for the first time while drafting the volume that has to overcome it.

Fix it with assignment, not vigilance. Every active contract gets a named responder, and that responder holds a pre-built response package so a reply is an assembly task rather than a memory exercise.

Response package checklist:

  1. Factor-by-factor claim, written to the specific evaluation language rather than to the overall rating
  2. Supporting artifact for each claim, attached, not referenced
  3. System of record named for every metric cited
  4. Date of each artifact and the metric definition behind it
  5. Approving program manager identified by name, with sign-off captured

Scenario. A program manager on a five-year sustainment order gets a CPARS notification showing a Satisfactory on Schedule, with narrative language citing two late deliverables. Because the program's delivery metrics, customer correspondence, and modification history already sit in a governed library keyed to the order number, the response takes an afternoon: both late deliverables trace to government-directed scope changes documented in Mod 0003 and Mod 0005, with the revised dates in the contract file. The response goes in within the review window with the modification record attached. That is a different outcome than the same manager writing from memory on day nine. The mechanics of turning that record into reusable narrative are covered in more detail in CPARS-aligned past performance automation.

Building the Platform Layer Without Building a Science Project

Four stages, in order. The sequencing matters more than the tooling.

Stage 1: Inventory. List every source of performance evidence for your top contracts by value. Assign a named owner, a join key, and a refresh cadence to each one. Done when every source on the list has all three filled in. Ingest nothing before this is complete, because an unowned feed with no key produces data you cannot reconcile and nobody maintains.

Stage 2: Ingestion. Stand up scheduled retrieval from SAM.gov data services [5] and the USAspending API [6], plus exports from internal program systems. Owner is the capture analyst or data engineer. Done when a monthly pull runs without manual intervention and writes to a single store with source and date stamps on every value. This replaces portal refreshing, which is the activity most capture teams mistake for research.

Stage 3: Attribute modeling. Map every record to the FAR 15.305 relevance attributes and your controlled scope vocabulary [2]. Owner is the capture manager, with program manager validation on metrics. Done when a relevance query returns a usable candidate list without manual filtering.

Stage 4: Proposal-asset layer. Produce pre-reviewed write-ups, quantified outcome statements, and transition and staffing summaries, each tagged with source, date, and originating contract. Owner is the proposal manager. Done when a recompete volume starts from reviewed content instead of a blank page.

That final layer is where the data investment pays off in proposal hours. An attribute-tagged library feeds compliance matrix automation directly, because requirement-to-evidence mapping stops being a manual hunt. It also feeds structured drafting, where writers assemble sections from approved, source-tagged components instead of recomposing the same past performance paragraph for the ninth time.

From Live Data to a Recompete Volume in Ten Days

Ten days is a realistic turn for an incumbent recompete when the data layer already exists. Here is the sequence.

Days one and two: run the relevance query and map candidate references to the Section M evaluation factors. Days three and four: build the gap list, showing open requirement, owner, missing evidence, next action, and due date. Days five and six: program manager validation on every quantified claim, with approver and date captured. Days seven through nine: drafting from reviewed components, with the gap list burned down in parallel. Day ten: review gate against the compliance matrix, not against general impressions.

Compare that to the manual cycle. Past performance selection consumes four days of debate rather than a two-hour query review. Metric verification happens at Red Team instead of before drafting, which means numbers get softened under time pressure. CPARS alignment never happens at all, because nobody pulled the evaluation narratives. The transition narrative gets written from the previous proposal, which was written from the one before that.

Archive the submission as a complete, retrievable package at the moment you submit: final volumes, pricing basis, submission confirmations, and the Section L and M compliance matrix. If the recompete does not go your way, FAR 15.506 governs post-award debriefing request timing and the categories of information agencies provide [10]. The limiting factor on a useful debriefing is almost never the agency. It is whether you can assemble your own file fast enough to draft questions that target the evaluation record.

Then feed every confirmed debriefing finding back into the library as a tagged lesson keyed to agency, evaluation factor, and vehicle. That is how the next pursuit inherits the correction instead of repeating it.

Frequently asked questions

What is recompete readiness?

Recompete readiness is the practice of governing incumbent performance data as a live dataset instead of reconstructing it when the solicitation posts. Every past performance record carries a named owner, a refresh date, and the relevance attributes FAR 15.305 directs evaluators to use, namely scope, magnitude of effort, and complexity [2], alongside the rating recorded in CPARS under FAR Subpart 42.15 [1]. The working test is whether a relevance query returns candidate references in minutes rather than days.

What data belongs in a recompete-ready past performance record?

The FAR 15.305 relevance attributes as structured fields: scope, magnitude of effort, complexity, and recency [2], plus agency, vehicle, contract or order number, unique entity identifier, CPARS rating by factor [1], and quantified outcome statements with source system and date on each one.

How often should a past performance library refresh?

Obligation data monthly, since public award and obligation records are available through the USAspending API on a schedule [6]. Delivery metrics monthly. Staffing quarterly. CPARS on a continuous monitoring basis, because the review and comment opportunity under FAR 42.1503 is date-bound [1].

Who owns CPARS monitoring?

A named responder per contract, usually the contract manager with program manager support. Shared ownership means nobody watches. The same ownership discipline runs across the whole pursuit, which is why this work belongs with capture managers rather than with whoever happens to be drafting that week.

Where does safeguarding apply?

FAR 52.204-21 sets basic safeguarding requirements for covered contractor information systems [3]. Apply access partitioning before opening the library to writers, and keep proprietary and source-selection-sensitive content separate from general reuse content.

Can competitors see our incumbent contract record?

Parts of it. Award, modification, and obligation records are published in machine-readable form through the USAspending API [6], and GSA publishes documented SAM.gov data services [5]. Secrecy is not the defense available to you. Governing that public record and joining it to your internal delivery metrics is, and it is the same discipline behind competitive intelligence gathered before the RFP posts.

Should we bid every recompete we hold?

Not automatically. An attribute-tagged record of your own performance is also the input to an honest bid decision, because obligation trend, scope drift, and factor-level CPARS history are exactly the evidence a fit and pWin assessment should weigh. Incumbency is a position to evaluate at the gate, not a reason to skip it.

What if the recompete does not go our way?

FAR 15.506 governs post-award debriefing request timing and the categories of information agencies provide [10]. Archive the submission package at the moment you submit so your own file is retrievable, then draft questions against the evaluation record rather than against general impressions. Feed every confirmed finding back into the library as a tagged lesson.

Your Next Two Weeks

Your one concrete action this week: run a source inventory on your three largest contracts by value. For each of the five source categories, name an owner and set a refresh date. That is a two-hour exercise that exposes exactly where your evidence is unowned.

Your one metric: percentage of past performance records carrying a named owner, a source tag, and a refresh date under 90 days. Start it this week even if the first number embarrasses you. Most teams start below 20 percent.

Go back to the opening scenario. Thirty days to the response, three weeks spent reconstructing deliverable dates. With a governed dataset, those three weeks become a two-day relevance query and a program manager validation pass, and the remaining time goes where it belongs: into the solution and the discriminators, not into archaeology.

References