Agency Case Study Verification Checklist

Use this checklist before treating an agency case study as selection evidence. Mark each item verified, supported, asserted, missing, or not transferable. An unverified result may still be a story about process, but it should not be presented as proof of future performance.

1. Define the claim

• [ ] The case study states what it is intended to prove.
• [ ] Client identity, service, period, market, and scope are clear.
• [ ] The agency’s role is separated from other suppliers and client actions.
• [ ] The buyer records what alternative evidence would be acceptable.

2. Verify the baseline

• [ ] Starting metric, comparison period, segment, and denominator are named.
• [ ] The source or export behind the metric is identified.
• [ ] Traffic, interaction, raw enquiry, accepted lead, opportunity, and revenue are separate.
• [ ] Seasonality, pricing, product, staffing, and platform changes are annotated.
• [ ] A small sample can be inspected without exposing unnecessary confidential data.

3. Verify the method and contribution

• [ ] The intervention, timing, owner, and dependency are described.
• [ ] The case distinguishes observation from causal attribution.
• [ ] Alternative explanations are acknowledged.
• [ ] The work can be compared with the buyer’s own constraints.
• [ ] The case identifies what would need to be tested in a pilot.

The Salesforce lead implementation guide is a useful reference for checking whether a case has separated lead capture from qualification and handoff. Local acceptance rules still govern.

4. Verify claims and permissions

• [ ] Client name, logo, quote, screenshot, data, and result have appropriate permission.
• [ ] Objective claims have evidence that supports the wording actually used.
• [ ] Dates, denominators, and conditions are not hidden by rounded language.
• [ ] Current platform, price, product, or policy claims have an expiry or review date.
• [ ] Legal or compliance review is assigned for material claims.

The FTC advertising FAQ provides general guidance on substantiation and responsibility for advertising claims. Escalate regulated or high-risk claims rather than relying on the checklist alone.

5. Check transfer limits

• [ ] Audience, offer, market, sales cycle, budget, and team are comparable.
• [ ] The buyer distinguishes a transferable mechanism from a copied result.
• [ ] Measurement and attribution conditions are comparable or will be tested.
• [ ] The agency can state what is not known from the case.
• [ ] A bounded capability sample or pilot is available when relevance is uncertain.

Google’s people-first content guidance can help evaluate content examples for purpose, usefulness, originality, and evidence. It is not a performance guarantee.

6. Test evidence custody and repeatability

• [ ] The provider can identify the source export, dashboard, brief, or approval record.
• [ ] The evidence includes date, property, filters, denominator, and relevant segment.
• [ ] A second reviewer can inspect the evidence without relying on a private sales explanation.
• [ ] The agency can describe a buyer-controlled pilot or diagnostic step.
• [ ] The first test has an owner, observation window, and stop condition.
• [ ] Scope, team, dates, and outcomes are consistent across page, proposal, call, and contract.

If the evidence cannot be inspected, keep the case as an assertion or process example. Do not upgrade its status because the design is polished.

7. Record the questions and answers

• [ ] Every unresolved claim has a written question.
• [ ] The provider’s answer is stored with the date and respondent.
• [ ] New evidence is linked to the exact claim it supports.
• [ ] Confidential material is redacted and access is limited.
• [ ] A reviewer records whether the answer changed the evidence status.

This record protects the buyer from memory drift during a fast shortlist. It also gives the agency a fair chance to clarify a case instead of being judged by a hidden standard.

When an answer changes the status, preserve both the original claim and the narrower approved wording. A case may move from asserted to process evidence without becoming performance proof. This distinction keeps a useful clarification from turning into an accidental guarantee.

8. Check shortlist fit and next step

• [ ] The case matches the buyer’s audience, offer, market, cycle, and operating constraints.
• [ ] The buyer can name the mechanism that might transfer, not just the result to copy.
• [ ] A closer case, capability sample, or bounded pilot is requested when fit is weak.
• [ ] The shortlist record states what the case does not prove.
• [ ] The final decision owner signs the evidence status before the case influences the score.

Ask the reviewer to write one sentence explaining the evidence boundary: what the case proves, what it only suggests, and what must be tested next. If the sentence cannot be written without broad words such as “success” or “growth,” keep the case in a bounded state.

Preserve that sentence with the shortlist decision and revisit it if the agency changes the proposed team, scope, or commercial promise.

The record should remain understandable after the shortlist meeting, without relying on a remembered sales conversation.

That is part of the evidence, not administrative overhead.

It gives the final decision owner a recoverable basis for approval before signing.

9. Choose the evidence gate

| Gate | Meaning | Action | | — | — | — | | verified and relevant | claim and conditions are inspectable | keep as one shortlist input | | process evidence | method is clear but result is bounded | request capability sample | | asserted | material proof is missing | do not use as performance proof | | not transferable | buyer conditions differ materially | request a closer case or pilot | | unsafe | rights or claims cannot be established | hold or exclude |

Store the record, source links, questions, reviewer, and date. A case-study checklist is complete when it narrows the next verification step and keeps an attractive story from becoming an unsupported forecast.