A pre-draft bid strategy test is a controlled attempt to disprove an already proposed strategy before writers turn it into persuasive copy. Each material claim is reduced to a buyer fact, an evaluated decision, an offered choice, a causal effect, applicable proof, delivery and price conditions, counter-evidence and a kill condition. The resulting record releases, narrows, suspends or rejects the claim. It does not create the strategy, predict competitors, choose an undisclosed evaluator preference or replace the organization’s bid and commercial authority.
A strategy workshop can produce consensus without producing a defensible bid. Once a polished theme enters the executive summary and section briefs, authors treat it as approved truth. Evidence is selected to support it, awkward facts become footnotes and an unpriced commitment spreads through several answers. The first serious challenge then arrives after hundreds of drafting hours, when changing the offer is expensive and withdrawing a headline feels like failure. The team has reviewed the wording, but it never defined what result would prove the strategy wrong.
Test the strategy as a set of defeasible claims, not as a presentation. Start with the strongest reasonable challenge to buyer relevance, evaluation route, causal logic, proof applicability, offer authority, economics and cross-bid consistency. Require the strategy owner to name the evidence that would change the decision. A claim that survives may enter a controlled response brief with its conditions attached. A claim that fails is narrowed, replaced or killed before prose gives it momentum. If a failed claim was central to the pursuit case, return the matter to the authorized bid decision rather than asking writers to compensate.
Test contract
Freeze what is being tested and what the result can authorize
A strategy cannot be tested against a moving package. Record the contracting authority, procurement, procedure, lot, bidder configuration, current notice, tender documents, amendments, clarifications, question set, award criteria and observation time. Then identify the proposed solution version, delivery plan, resource view, pricing basis, contract positions and evidence snapshot. If the review spans several lots or variants, open a separate test lane wherever buyer facts, evaluation, offer choices or economics differ. A result for one configuration says nothing about another unless the record proves the shared boundary.
State the decision the exercise may make. A claim test can authorize one proposition for named response locations. It cannot approve the whole submission, waive a mandatory condition, select a price or silently reverse a no-bid rule. Name the strategy owner, independent challenger, solution owner, finance owner, evidence owner and pursuit authority. One person may hold more than one role in a small team, but nobody should both originate a material claim and supply its only challenge. The AQuA Book separates analysis and assurance roles for a similar reason. That guidance is not a tender rule for suppliers; it is a useful control when an internal conclusion will drive expensive work.
Define proportional depth before the meeting. A theme affecting one low-weight narrative may need a short source and proof check. A claim that changes staffing, price, contract risk or several scored answers needs independent evidence inspection and owner approval. The test should finish early enough to change the offer. A long review held after the writing plan is locked is still late, even if its minutes are excellent.
Northport Flood Partnership, a fictional authority used throughout this edition, is buying a five-year tidal-barrier inspection and maintenance service. The proposed strategy says remote condition screening will reduce emergency closures and technician exposure. The test baseline fixes Lot 2, the issued maintenance method question, the service-continuity criterion, the bidder’s revision B operating model and price workbook P3. It does not assume that the claim is true merely because the technology team demonstrated a survey vehicle at kickoff.
| Field | Recorded answer | Stop condition |
|---|---|---|
| Boundary | Procurement, lot, stage and bidder configuration | Applicable package is unresolved |
| Buyer baseline | Current sources and evaluation model | Material source is missing or disputed |
| Offer baseline | Solution, delivery, price and contract versions | Versions do not describe one offer |
| Decision right | What this review may release | Reviewer lacks the required authority |
| Challenge role | Named reviewer independent of claim origin | No credible challenge is available |
| Expiry | Events that reopen the result | No owner will monitor the triggers |
Claim model
Rewrite the strategy so that it can lose
Break every proposed theme into six objects. First comes the buyer premise: a fact or bounded inference from the current record. Second is the evaluation route: the criterion, question, descriptor or permitted decision where the fact can matter. Third is the offer choice: the method, resource, control or commitment that the bidder will actually provide. Fourth is the mechanism that connects the choice to a consequence. Fifth is the proof proposition. Sixth is the condition that limits where the reasoning holds. Keep the objects separate even when the final theme later compresses them into one sentence.
Write each proposition so that an observation can change its state. “Remote inspections improve resilience” cannot lose because improve has no defined object, comparison or boundary. Northport rewrites it as: “For tidal gates accessible to the proposed vehicle, monthly remote screening under operating model B identifies the listed defect classes between planned engineering visits; maintenance records from the bidder’s comparable estuary contract show that this route shortened the median time from observable defect to triage during the stated period.” This version exposes access, defect coverage, cadence, model, comparator, metric and evidence period. Any one can fail.
Keep buyer fact, bidder inference and proposed commitment visually distinct. A specification that lists unplanned closures establishes that the events exist; it may not establish their cause or the buyer’s private priority. A high weighting shows relative importance only within the published model. It does not prove that one feature will receive a particular score. FAR 15.304 and the UK guidance on assessing competitive tenders describe evaluation against stated factors in their respective regimes. The practical supplier discipline is narrower: never use an internal story to enlarge the buyer’s disclosed decision.
Attach defeat evidence before collecting more support. The claim may fail if the tender prohibits the access method, the criterion cannot credit reduced triage time, the vehicle misses a material defect class, the cited contract used different assets, the submitted team lacks the trained operator, or the subscription and standby cost are not priced. A claim with no imaginable failure condition is not ready for review. It is advertising copy waiting for confirmation.
Buyer relevance
Make the claim earn a place in the published decision
Start with a cold reading of the buyer record. Locate the concern, desired outcome, constraint or tradeoff without using the strategy deck. Preserve exact wording and source version, then mark the interpretation needed to connect it to the claim. A requirement for twenty-four-hour fault intake supports a need for availability. It does not by itself prove dissatisfaction with the incumbent or a preference for automation. Preliminary market engagement material can supply context when it belongs to the procurement record, but it should not override the released competition or private information-sharing rules.
Next prove the route to evaluation. Map the claim to the exact question, requested component, award criterion, subcriterion, pass condition or presentation stage that can receive it. Read any score descriptors and limits. If service continuity is assessed through mobilization, incident response and recovery evidence, a remote-screening theme must show how it answers one of those decisions. A benefit that sits outside the methodology may still be useful in delivery, but writers should not sacrifice requested content to promote it. Under section 23 of the UK Procurement Act 2023, award criteria and the assessment methodology have defined transparency and subject-matter requirements. Other regimes differ, so the live tender and applicable rules control.
Test materiality rather than merely locating a keyword. Ask what evaluator conclusion could change if the claim were proved, and what the answer would say if the claim were removed. If the only difference is a more modern-sounding introduction, kill the theme. If the claim supplies a requested method, closes a named risk or demonstrates a higher published distinction, record that path. Do not translate criterion weight into a predicted score. The test establishes relevance and a supportable response use, not the evaluator’s future judgement.
Northport finds two different results. Remote screening is relevant to the maintenance-method question because the buyer asks how latent defects will be identified between planned visits. The claim that it will reduce technician exposure has no requested measure, safety method or evidence path in that answer. The team keeps the safety control in the operational risk plan but removes it from the central win claim until a legitimate response location and proof are found.
| Test | Evidence required | Fail result |
|---|---|---|
| Buyer premise | Current buyer-authored fact or bounded inference | Remove invented concern |
| Assessment route | Criterion, question or stated decision | Do not displace scored content |
| Requested object | Method, proof, plan or commitment the buyer can inspect | Rewrite or relocate claim |
| Materiality | Evaluator conclusion that proof could change | Kill decorative theme |
| Boundary | Unknowns and conditions kept visible | Prevent unconditional release |
Causal and proof test
Try the strongest alternative explanation before accepting the evidence
A relevant claim can still be wrong. Draw the causal chain from offered choice through operating behavior to buyer consequence. At each link, ask what else could explain the cited result and what the offer must contain for the mechanism to work. Faster defect triage might come from a larger duty team rather than remote screening. Fewer closures might reflect newer assets, a milder season or a different maintenance threshold. The review need not prove a scientific causal effect, but it must stop a correlation, anecdote or demonstration from becoming a certain contract outcome.
Inspect the evidence object, not its internal label. Record who issued it, which legal entity and team performed the work, what assets and geography it covered, the service model, period, measure, denominator, baseline, exclusions and permission to disclose. FAR 15.305 expressly identifies currency, relevance, source and context in its treatment of past performance for covered US acquisitions. That provision is not a universal supplier evidence standard, but the questions travel well. Evidence from an urban drainage survey may establish vehicle capability while remaining weak evidence for operation around tidal gates in salt water.
Grade each proof proposition by what it establishes. Direct support matches the offered entity, relevant scope, operating conditions and measure. Partial support establishes only a component, such as sensor detection under laboratory conditions. Transferable support requires a stated reason why differences do not break the mechanism. Contradictory evidence weakens the claim. Missing evidence stays missing even when an owner promises to find it later. Do not average several weak artifacts into strong proof. Their shared gap may be the very condition that matters.
Northport’s case study covers remotely operated seawall surveys, not moving tidal gates. It reports inspection throughput but no defect-to-triage time. The demonstration proves that the proposed vehicle can transmit images in sheltered water. It does not prove coverage of gate recesses, availability in the specified current range or the claimed closure outcome. The team therefore rejects the measured-outcome wording. It retains a narrower method claim, conditional on access testing and supported by the vehicle specification, operator record and a planned acceptance test.
| Evidence dimension | Question | Possible state |
|---|---|---|
| Authority | Who created and controls the record? | Verified, bounded or unknown |
| Entity and role | Who performed the evidenced work? | Same, relied-on or different |
| Scope and conditions | Do assets, users, geography and service model match? | Comparable, transferable or not comparable |
| Measure | Does the record prove the claimed variable and baseline? | Direct, proxy or absent |
| Period | Is the observation current for the capability? | Current, explainable or stale |
| Disclosure | May the evidence and attribution enter the bid? | Approved, redact, summarize or withhold |
Offer reality
Check whether the submitted offer can carry the claim
Move from evidence of past capability to the proposed contract. Locate the choice in the current solution baseline, work breakdown, staffing plan, partner commitments, bill of materials, schedule, service levels, price model and contract response. Record the owner authorized to make each promise. A product feature that exists in a demonstration environment is not part of the offer until its configuration, license, deployment, support and acceptance path are included. A partner’s marketing page is not a capacity commitment. A workshop preference is not approval.
Run the economics at the same boundary as the claim. Identify incremental equipment, people, training, mobilization, data, assurance, contingency and third-party costs. Check where they appear in the price schedule and whether volumes, indexation, taxes, currencies or buyer dependencies change the result. Use scenarios for unresolved inputs and name the switching condition at which the choice stops being viable. The Green Book 2026 discusses sensitivity and switching values for public appraisal, not supplier pricing. Its useful methodological lesson here is to expose which assumption changes the decision instead of hiding uncertainty inside one favored total.
Test operational capacity and timing. Confirm that named people are available to the bidding entity, facilities can be ready by the contractual milestone, certifications cover the proposed activity and subcontract arrangements match the submission. Distinguish a curable condition with an owner and date from an unavailable capability. If a central claim depends on buyer access, data, approvals or volumes, show the dependency wherever the claim appears and confirm the tender permits the proposed treatment. Do not turn a dependency into an implicit buyer promise.
Northport discovers that operating model B includes one vehicle but no standby unit, specialist training or annual software subscription. The price lead also used planned engineering visits as the labor baseline, so the claimed between-visit triage service is unpriced. The strategy owner cannot release the original claim. The solution and finance owners may add a bounded service and reprice it, or the team may revert to a conventional inspection position. Writers cannot solve the gap by calling the unpriced concept “included.”
Challenge session
Run a decision review that rewards disconfirming evidence
Give the challenger the frozen sources, proposition records and evidence objects before the session, but not the polished strategy narrative as the primary input. Ask them to reconstruct the claimed buyer decision and offer consequence. The strategy owner then shows where the reconstruction differs. This catches reasoning that exists only in the workshop history. It also prevents a confident presenter from substituting fluency for an inspectable chain.
For each proposition, lead with the strongest plausible failure. The challenger can test an alternative buyer interpretation, a different cause for the past result, a comparable competitor response, an offer dependency, a cost boundary or contradictory evidence. The claim owner must answer with a source, approved choice or explicit uncertainty. Silence is not failure by itself, but it cannot become support. Record new work only where a named artifact could close the gap before the drafting decision. Avoid a loose list of actions that postpones every hard conclusion.
Use state changes rather than votes. The reviewer recommends a state and cites the reason. The owner with the relevant authority decides any residual exposure: evidence owners decide disclosure, solution owners decide the method, finance decides the approved economics, legal or commercial reviewers decide their terms, and the pursuit authority decides whether a central failure changes the bid. Majority enthusiasm cannot grant authority that the room does not possess.
Keep the session short enough to force decisions, then allow evidence inspection outside it. A two-hour meeting cannot authenticate every record, while an open-ended “red team” can consume the time it was meant to protect. Timebox the oral challenge, assign exact verification tasks and reconvene only the affected propositions. The review closes when each material claim has a state, owner, evidence basis, restriction and expiry event.
Decision rules
Kill the claim at the level where the evidence fails
Define kill conditions before the support case is presented. A claim is rejected when a current buyer source makes it irrelevant or impermissible, a mandatory offer condition cannot be met, the mechanism is contradicted, required proof is unavailable, the commitment lacks authority, the economics fall outside the approved boundary or the claim conflicts with another controlling bid position. “Senior team does not like it” is not a kill condition. Neither is one reviewer’s preferred style. Conditions should identify an observable state and the decision it changes.
Apply the failure at the narrowest valid level. Weak outcome evidence may kill a quantified benefit while leaving a factual method description. An unavailable partner may kill one delivery route while leaving the buyer problem and a different offer choice intact. No evaluation route kills use as a win theme but does not necessarily remove the feature from the technical plan. Preserve the reason and the retired wording so it cannot reappear from an old slide or content library.
Some failures reach beyond a theme. If the strategy depended on meeting a threshold, closing a central capability gap, sustaining the submitted price or differentiating an otherwise weak offer, its failure may invalidate the selected configuration or the original pursuit case. Route that consequence to the owner of the offer or bid decision. The stop-a-bid process owns withdrawal after investment; this test supplies evidence to that decision but does not make it automatically.
Allow conditional release only when the condition is bounded and governable. Name the missing artifact, accountable owner, latest date, dependent sections and fallback wording. Prevent final drafting or claim use until the condition closes when the exposure is material. A vague state such as “approved subject to evidence” merely moves the risk into the writing queue.
| State | Meaning | Drafting consequence |
|---|---|---|
| Released | All required tests pass at the recorded boundary | Use only in approved locations |
| Released with condition | Bounded gap has owner, date and fallback | Restrict use until stated control closes |
| Narrowed | Only part of the proposition survives | Issue replacement wording and proof |
| Suspended | A material test remains unresolved | No persuasive drafting from the claim |
| Rejected | A kill condition is met | Withdraw from every brief and artifact |
| Escalated | Failure changes offer or pursuit viability | Return to authorized decision owner |
Drafting release
Give writers a tested reasoning chain, not a slogan
The released brief should contain the buyer source, assessed decision, exact offer choice, permitted claim, causal explanation, proof records, conditions, prohibited extensions, intended response locations and review question. Include the current IDs or locators for solution, price and contract decisions. Writers may adapt syntax to answer each question, but they may not strengthen the claim, remove a condition or transfer evidence to a different scope. A central headline is optional. The reasoning and control are not.
Northport ultimately releases a narrower position: revision C provides remote visual screening for the named accessible gate surfaces between planned visits, with suspected defects triaged by the duty engineer under the stated response process. The brief cites the vehicle capability record and operator competence evidence, describes the acceptance test and forbids claims about reduced closures until contract-specific data exist. Finance adds the subscription, training and standby arrangement to P4. The strategy has less drama than the kickoff theme and far more chance of surviving evaluation and delivery.
Reconcile every use after drafting. Search headings, diagrams, executive summary, method answers, implementation plan, risk register, price narrative, assumptions and contract schedules for the claim and its close variants. A condition that appears only in the source brief will not protect an unconditional promise in the submitted page. The criterion strategy process owns how released claims combine with other response decisions, while the blank-page and final readiness reviews own the later rendered response. This dossier ends at controlled release and change monitoring.
Automation can inventory source links, compare versions, locate repeated claims, surface missing owners and test whether approved evidence is cited. It can also propose counter-questions. People retain interpretation of buyer intent, authority over the offer, evidence-release decisions, commercial acceptance and the pursuit choice. Require every automated finding to point back to the controlled object it inspected. Do not let a generated confidence score replace the recorded reason for release or rejection.
Revalidation
Reopen the test when the facts beneath the strategy move
Every result needs an expiry rule. Reopen affected claims after an amendment, clarification, criterion change, response-format change, solution decision, partner withdrawal, capacity update, evidence correction, pricing revision, contract position or new contradictory fact. Do not rerun the whole workshop when one input changes. Use dependencies to identify which propositions and briefs inherit the change, then preserve the prior state for audit.
Monitor unresolved conditions on their own dates. If a proof letter was due before the authoring start and does not arrive, apply the recorded fallback instead of extending the condition informally. If a buyer answer changes the premise, update the source object first and reassess relevance before debating wording. If a new price removes the economic margin, the finance owner reopens the offer test even when all prose is finished. A release is current only while its evidence and decisions remain current.
Close the pre-draft record when all material claims are released, narrowed or rejected, each central failure has reached the proper decision owner and every writer brief reflects the current state. Report open conditions separately from passed claims. The completion measure is not the number of green themes. It is whether the team knows which reasoning it may use, which reasoning it must not use and which events change that answer.
After submission, retain the record with the submitted response and later award evidence. A debrief may support, weaken or leave the hypotheses unresolved; it rarely proves the entire internal strategy. Compare specific evaluator findings with the propositions that were tested. Feed any new evidence into the separate win-loss process. Do not rewrite the old test so that it appears to have predicted the result.
What good looks like
Useful outcomes from test bid strategy before drafting
- Every material strategy claim is stated in a form that evidence can support or defeat.
- Buyer relevance is traced to the current procurement record and a permitted evaluation route.
- The proposed offer choice exists in solution, delivery, price and contract positions.
- Proof is tested for entity, scope, period, conditions and permission to disclose.
- Counter-evidence and unresolved dependencies remain visible beside favorable evidence.
- Each claim has a named kill condition, decision owner and expiry trigger.
- Writers receive only released claims, proof boundaries and approved response uses.
- A failed central premise reopens the pursuit or offer decision before drafting consumes more capacity.
Operating model
How to run the work
- 01
Freeze the test baseline
Identify the procurement, lot, stage, current documents, evaluation model, proposed offer, price basis, evidence snapshot and decision owners that the test covers.
- 02
Atomize the strategy
Split each theme into buyer premise, offered choice, claimed consequence, proof proposition, conditions and intended response locations.
- 03
Name defeat evidence
Before reviewing support, state the observation that would make each proposition irrelevant, infeasible, uneconomic, unprovable or contradictory.
- 04
Test the evaluation route
Verify that the current criteria, questions and methodology can credit the claimed advantage without inventing a private buyer preference.
- 05
Challenge mechanism and proof
Try credible alternative explanations, inspect proof applicability and test whether the proposed choice can cause the stated contract consequence.
- 06
Reconcile the actual offer
Check authority, capacity, dependencies, cost, price, contract treatment and every other answer that would inherit the claim.
- 07
Decide claim state
Release, release with conditions, narrow, suspend or reject each claim and route a failed central premise to the authorized pursuit owner.
- 08
Issue and monitor briefs
Give writers the released reasoning, proof and limits, then reopen the record when a source, offer, price, evidence or capacity input changes.
Evaluation
Questions that change the decision
- Which exact version of the procurement and proposed offer is being tested?
- What buyer fact and assessed decision make this claim relevant?
- What offered choice is materially different from the minimum compliant position?
- What causal link connects that choice to the claimed buyer consequence?
- What evidence would support the link under comparable conditions?
- What credible observation would defeat or materially narrow the claim?
- Can the offered entity deliver and authorize the promise at the submitted price?
- Which answers, schedules and contract positions must remain consistent with it?
- Is the residual uncertainty acceptable to the owner with authority over the exposure?
- Does a failed claim affect only one theme, the selected offer or the decision to bid?
Failure modes
Where teams lose control
The review tests presentation quality instead of the underlying proposition.
The team searches only for support after agreeing the preferred story.
A documented buyer issue has no route through the published assessment method.
A plausible mechanism is described as a measured outcome without comparable proof.
A case study belongs to another entity, role, scope, period or operating condition.
A solution owner likes the concept but has not approved the offered commitment.
A benefit depends on cost or capacity absent from the price and delivery baseline.
A condition is removed when a claim is shortened for a heading or summary.
An amendment or internal design change leaves a previously released claim in circulation.
A failed central premise is hidden so that drafting can continue on schedule.
Measurement
Measure the finished job
Measure the completed workflow, including review effort and exceptions. Output volume on its own is not evidence of a better process.
- material claims with explicit defeat evidence before support review
- claims linked to an exact buyer source and assessable response location
- claimed effects with an inspected causal mechanism and applicable proof
- offer commitments reconciled across solution, capacity, price and contract
- claims narrowed or rejected before author assignment
- open conditions with an owner, deadline and drafting restriction
- writer briefs containing only released claims and current evidence
- late changes that reopen every dependent claim record
- drafting hours avoided after early strategy failure
Questions
Common questions
Is this the same as building a bid strategy?
No. Strategy construction chooses buyer problems, offer positions, proof priorities and tradeoffs. This test starts after a proposed strategy exists and tries to defeat its material claims before they govern drafting.
Is a pre-draft strategy test just an early red team?
Not if the red team mainly reviews persuasion or a draft. This test examines propositions, counter-evidence, offer authority and kill conditions before prose. Later reviews still need to assess the rendered response.
Does every claim need quantitative proof?
No. The proof must match the claim. A method claim may need design, competence and control evidence. A quantified outcome needs a defined measure, baseline, period and applicable observation.
What is a useful kill condition?
It is an observable event that changes a decision, such as no permitted evaluation route, an unavailable mandatory resource, contradictory evidence, an unapproved commitment or economics outside the accepted boundary.
Can a claim pass with an open evidence gap?
Only as a bounded conditional state with an owner, due date, restricted uses and fallback. Material persuasive use stays blocked until the required evidence or decision exists.
Who should challenge the strategy?
Use someone able to read the procurement and question the offer who did not originate the claim. Relevant owners then decide evidence, solution, finance, contract and pursuit issues within their own authority.
Should writers attend the strategy test?
They may help expose ambiguity and intended uses, but they should not be asked to cure missing buyer evidence, approve the offer or turn a failed proposition into persuasive language.
When must a passed strategy be tested again?
Reopen affected claims when buyer sources, evaluation, the offer, evidence, capacity, price, contract positions or relevant contrary facts change. Use recorded dependencies to limit the retest.
Sources
Primary references
- Guidance on assessing competitive tenders under the Procurement Act 2023 Government Commercial Function
- Official Procurement Act learning module on assessment and award Government Commercial Function
- Guidance on preliminary market engagement Government Commercial Function
- Procurement Act 2023 section 23 on award criteria The National Archives
- Procurement Act 2023 section 24 on refining award criteria The National Archives
- FAR 15.201 on exchanges before receipt of proposals Acquisition.gov
- FAR 15.304 on evaluation factors and relative importance Acquisition.gov
- FAR 15.305 on proposal evaluation and past performance context Acquisition.gov
- FAR 15.306 on exchanges with offerors after receipt of proposals Acquisition.gov
- Directive 2014/24/EU, consolidated English text EUR-Lex
- World Bank Rated Criteria guidance for suppliers World Bank
- World Bank guidance on evaluating bids and proposals with rated criteria World Bank
- World Bank Procurement Regulations, seventh edition World Bank
- ADB merit point criteria procurement guidance note Asian Development Bank
- ADB merit point criteria resource page Asian Development Bank
- UNCITRAL Model Law on Public Procurement United Nations Commission on International Trade Law
- New Zealand Government guidance on evaluating responses New Zealand Government Procurement
- AQuA Book guidance on analytical quality assurance UK Government Analysis Function
- The Green Book 2026 on sensitivity and switching values HM Treasury
- The Orange Book on challenge, evidence and risk decisions HM Treasury
About the author
Tony Kim
Tony writes about applied AI, dependable product engineering and the systems that turn complex response work into controlled delivery.
Zelius
Managed tender intelligence and bid execution for teams that want the commercial outcome.
Suppliers, founders and commercial teams pursuing public or private opportunities. Start with the workflow, constraints and evidence you already have.
