---
title: "Choose the proof points an evaluator can actually score"
description: "Select proof for the exact assessed claim, show why it applies to this offer, and place enough context beside the answer for an evaluator to use it."
canonical: "https://zephior.com/insights/choose-proof-points-for-an-evaluator"
last-updated: 2026-09-02
---

# Choose the proof points an evaluator can actually score

> Select proof for the exact assessed claim, show why it applies to this offer, and place enough context beside the answer for an evaluator to use it.

By [Tony Kim](https://zephior.com/authors/tony-kim). Published 2026-09-02; updated 2026-09-02. 12 minute read.

## Definition

A tender proof point is a compact, inspectable piece of support for one material response claim. It identifies what happened or exists, who or what it covers, the relevant scope and period, how it was established, and why it applies to the offered solution. Choosing proof points means ranking candidate evidence for the evaluator’s exact decision and response constraints. It is not the broader work of collecting an evidence library or defining organization-wide assurance levels.

## Problem

Teams often equate proof with a list of customer logos, years in business, certification badges and large numbers. The facts may be accurate yet fail to support the claim being evaluated. A global project does not prove local mobilization. A certificate does not prove every control in the proposed service. A percentage arrives without population or period. Writers include six weak examples because selecting one requires judgement, then force the evaluator to infer relevance. Stronger, smaller evidence stays buried in a technical record.

## Point of view

Select proof from the evaluator’s decision backward. First atomize the claim and identify what must be believed. Then compare candidates for directness, applicability, credibility, recency, distinct contribution and usability in the permitted format. Prefer one complete relevant proof point over several impressive but adjacent facts. Put the essential context beside the claim, state material limits and use a precise reference when supporting material is allowed. If no evidence fits, narrow the claim or change the response rather than asking the evaluator to bridge the gap.

## Separate the claim before searching for an impressive example

Start with the criterion, response question and intended answer. Underline every statement that asks the evaluator to believe something material. “Our certified method delivered a 30 percent faster deployment and will protect the buyer’s launch date” contains several propositions: the method has a named certification, the offered scope falls within it, a deployment was measured, the comparison is defined, the method contributed to the result and the future plan can meet this deadline. One logo or case study cannot automatically prove the chain.

Write one proof job for each necessary proposition. A proof job describes the fact to establish and the context that makes it relevant: “Show that the proposed mobilization lead has directed a transition of comparable service count, deadline pressure and continuity requirement.” It is more useful than “add a case study.” The first instruction guides selection; the second encourages the nearest polished document whether or not it answers the evaluator’s decision.

Classify the claim. A current capability needs evidence that the function exists in the offered configuration. Past performance needs the bidder’s role, comparable scope and period. A quantitative result needs population, baseline, calculation and conditions. A causal statement needs support for the mechanism, not only two correlated facts. A future commitment needs authorized resources, design, feasibility and consistent contractual wording. The proof format follows the proposition.

Keep the proof burden proportional to the claim and its evaluation consequence. Routine context may need a controlled current source. A central claim about regulated compliance, service availability, price reduction or delivery date needs stronger and often more independent support. This article chooses among available candidates for an answer; the separate evidence-hierarchy process should already determine the minimum assurance expected for high-consequence claims.

**Proof job by claim type**

| Claim | What must be established | Weak substitute |
| --- | --- | --- |
| Capability | Function exists in offered scope | Generic product page |
| Experience | Entity and role in comparable work | Customer logo |
| Result | Defined measure, population and period | Percentage alone |
| Cause | Credible mechanism and alternatives | Before and after only |
| Assurance | Issuer, entity, scope and validity | Badge image |
| Commitment | Authority, resources and feasibility | Historic aspiration |

## Choose the candidate that proves this offer, not the one with the best name

Create a shortlist and describe each candidate before ranking it. Record source owner, issuing party, bidding entity, delivery role, product or service, geography, customer type, scale, complexity, period, method, result, conditions, current validity and permission to disclose. Include the exact passage or data field. A filename is not a description. This record makes mismatches visible before a writer turns them into fluent but misleading prose.

Score candidates qualitatively on six questions. Directness asks whether the item proves the exact proposition. Applicability compares entity, role, scope and conditions with the offer. Credibility covers source authority, control and independence. Recency tests whether the fact remains valid. Distinct contribution asks what gap the item closes beyond existing proof. Usability checks whether it can be cited, attached or summarized under the response rules. Do not add the scores into false precision; use them to expose the selection tradeoff.

FAR 15.305 discusses past performance through currency, relevance, source, context and general trends, and notes that relevance determinations matter. The rule applies in its own federal acquisition context, but the questions are valuable for proposal evidence generally. A recent small project may prove the named person and method. An older complex project may prove scale. If both are allowed and each closes a different gap, use both and state their different jobs. Do not present quantity as independent confirmation.

Prefer the smallest sufficient set. Three references with the same role, method and marketing source often add less than one controlled project record plus one independent assurance artifact. Remove prestige facts that make no assessed proposition more believable. When the strongest candidate is only partially relevant, state the difference and limit the claim. The evaluator should never need to discover that the supplier was a subcontractor after reading language that implies prime responsibility.

- Directness: proves the proposition rather than a neighboring fact.
- Applicability: matches entity, role, scope, scale and conditions.
- Credibility: comes from an accountable and inspectable source.
- Recency: remains valid for the offered decision.
- Distinct contribution: closes a gap not already covered.
- Usability: can be considered in the required format and location.

## Put enough context beside the claim for the evaluator to use it

Package the selected point as claim, evidence, context and relevance. State the result or fact, identify the source and subject, give the dimensions needed to interpret it, then explain why it supports the offered method. “Reduced handling time by 28 percent” is incomplete. A usable proof point might identify the performing entity, workflow population, baseline and measurement period, what changed, the controlled record that contains the result and the material similarities and differences to the proposed scope.

Place the essential proof in the assessed answer when rules permit. An exact cross-reference can lead to supporting material, but “see case study” forces the evaluator to reconstruct the argument. Include the local conclusion and the context necessary for fair interpretation, then reference a stable final identifier such as “Reference 2, section 3.1, verified service metrics.” Check page, attachment and portal rules. Never rely on a cloud link, hidden comment or evidence location the buyer did not authorize.

Compress without stripping qualifiers. Keep entity, role, measure, period and comparison basis when they determine meaning. Remove decorative background, company history and details unrelated to the proof job. A table often works for several references because it exposes comparability rather than hiding it in narrative. Use plain labels that answer the evaluator’s scanning questions. ONS guidance on structuring content supports front-loading and descriptive headings; apply that to make evidence findable rather than promotional.

For confidential evidence, follow the tender’s process. Determine whether a redacted artifact, customer confirmation, auditor statement, controlled viewing route or permitted summary can establish the fact. State what the evaluator is actually receiving. Do not say “evidence available on request” as though unavailable support has already been examined. Secure disclosure approval for customer identity, personal data and contract information before insertion.

**Compact proof-point structure**

| Element | Evaluator question | Minimum content |
| --- | --- | --- |
| Claim | What should I believe? | Bounded proposition |
| Subject | Who or what produced it? | Entity, role and service |
| Context | Under which conditions? | Scale, period and baseline |
| Source | How was it established? | Record, issuer and method |
| Relevance | Why does it apply here? | Material similarities and differences |
| Reference | Where can I inspect it? | Permitted final identifier |

## Remove proof that asks the evaluator to make the missing argument

Give a reviewer only the buyer-visible answer and attachments. Ask them to restate what each proof point establishes, the scope it covers and the claim it does not establish. If the reviewer infers more than the source, the prose is overbroad. If they cannot explain relevance without private context, add the missing link or choose another item. The review should challenge source and claim together, not fact-check a citation in isolation.

Test each additional item by removal. If deleting it changes nothing the evaluator can conclude, it is clutter. If two items share a common origin, do not call them corroboration. If a proof point only repeats a mandatory credential already supplied elsewhere, reference it precisely rather than consuming scored space unless the criterion explicitly evaluates it. Evidence competes for attention; abundance can make the strongest support harder to find.

Revalidate dates, scope and offered configuration at final review. Certificates expire, named staff change, product editions differ and case-study permission can be withdrawn. Confirm every attachment opens and every reference resolves after PDF generation or portal upload. Inspect truncation, footnotes, table headers and page breaks. A qualifier separated from its number can change the claim even when the source text was correct.

When no candidate meets the proof job, make an explicit decision. Obtain a stronger record if time allows, change the offer to one that can be supported, label a forecast with its assumptions, narrow the proposition or delete it. Escalate a material residual claim to the authorized owner. The selection process succeeds when the evaluator receives a smaller, truer and more usable evidence case, not when every credential reaches the page.

- Ask a fresh reviewer what the proof actually establishes.
- Delete items that add no distinct conclusion.
- Recheck entity, role, validity and disclosure permission.
- Open every referenced artifact in the final package.
- Narrow or remove claims that no available evidence can support.

## Useful outcomes

- Every material response claim has a stated proof job.
- Candidate proof is compared against the exact criterion and offered scope.
- Prestige cannot compensate for a mismatch in entity, role, service or conditions.
- The selected item includes enough context to be interpreted correctly.
- Each additional proof point contributes something distinct rather than repeating one origin.
- Evidence appears where the evaluation rules permit it to influence the decision.
- Unsupported claims are narrowed, qualified, changed or removed.
- The final rendered response preserves proof identifiers and meaning.

## Workflow

1. **Define the proof job.** Break the answer into factual, quantitative, experiential, causal and future claims, then state what an evaluator would need to verify for each one.
2. **Build a candidate shortlist.** Find records that directly address the proposition and record entity, role, scope, period, conditions, source and disclosure permission.
3. **Rank for this decision.** Compare directness, applicability, credibility, recency, distinct contribution and response usability rather than source prestige alone.
4. **Package the proof point.** Write claim, evidence, context, relevance and limitation in a compact unit with an exact attachment or record reference where allowed.
5. **Test buyer-visible use.** Ask a reviewer to interpret the proof from the rendered response without private background and confirm it supports no more than the claim states.

## Key decisions

- What exact proposition must the evaluator believe to credit this answer?
- Is the claim about present capability, past performance, measured outcome, comparison or future commitment?
- Does the candidate cover the same bidding entity, role, service and conditions?
- Is the record direct evidence or support for a nearby fact?
- What context is necessary to interpret the number, result or certificate?
- Does a second item close a different gap or merely repeat the first source?
- Can the evaluator access and consider the evidence in the permitted response path?
- Must the claim be narrowed when the best proof is only partially applicable?

## Risks

- A famous client name substitutes for comparable scope and bidder role.
- A large percentage appears without baseline, denominator, period or method.
- A certificate is applied outside the named entity or assurance boundary.
- Several documents derived from one source appear to corroborate one another.
- A future service promise is supported only by a past marketing statement.
- All context is moved to an appendix the evaluator may not consider.
- Confidentiality or disclosure limits are ignored.
- The source is valid at draft time but expired or superseded before submission.
- Formatting truncates the proof qualifier while preserving the bold claim.

## Metrics

- material claims with an explicit proof job
- selected items matching offered entity, role and scope
- quantitative proofs with population, period and method
- proof points carrying disclosure approval and valid dates
- multiple items with a recorded distinct contribution
- claims narrowed after applicability review
- proof references resolving in the final response
- reviewer inferences required to understand relevance
- prestige-only facts removed from scored answers

## Frequently asked questions

### How many proof points should support one tender answer?

Use the smallest set that establishes the material claim and closes distinct credibility or applicability gaps. More examples are not stronger when they repeat one source or prove the same nearby fact.

### Is a well-known customer always the strongest reference?

No. The bidder’s role, comparable scope, conditions, recency and inspectable result matter more than name recognition. A less famous but directly comparable record may be far more useful.

### Can a certification prove the proposed service is compliant?

Only within its issuer, entity, scope, version and validity. It proves the certification described, not automatically every control, configuration or contractual requirement in the offer.

### What should happen when the proof is only partly relevant?

State the material difference, narrow the claim to what the record establishes and add complementary evidence only if it closes the specific gap. Never hide a role or scope mismatch.


## Primary sources

- [FAR 15.305 Proposal Evaluation](https://www.acquisition.gov/far/15.305), Acquisition.gov
- [FAR 15.304 Evaluation Factors and Significant Subfactors](https://www.acquisition.gov/far/15.304), Acquisition.gov
- [Guidance: Assessing Competitive Tenders](https://www.gov.uk/government/publications/procurement-act-2023-guidance-documents-procure-phase/assessing-competitive-tenders-html), UK Cabinet Office
- [Writing and Editing: Structuring Content](https://service-manual.ons.gov.uk/content/writing-for-users/structuring-content), UK Office for National Statistics


## Related articles

- [How old is too old for proposal evidence?](https://zephior.com/insights/decide-when-rfp-evidence-is-too-old)
- [How to anonymize a case study without destroying its value](https://zephior.com/insights/anonymize-case-study-evidence-safely)
- [Build win themes that survive evaluator scrutiny](https://zephior.com/insights/build-evidence-backed-win-themes)
- [How to balance compliance and persuasion in a tender response](https://zephior.com/insights/balance-compliance-and-persuasion-in-a-bid)