Practical guide
AI competitor brief verification checklist
A competitor brief can look polished while mixing unlike products, stale prices, unattributed assertions, and calculations that no reviewer can reproduce. This checklist is for the final review of a completed brief. It tests whether a decision-maker can trace the comparison from question to evidence and understand what remains uncertain. It does not choose competitors or write the brief. It also.
A competitor brief can look polished while mixing unlike products, stale prices, unattributed assertions, and calculations that no reviewer can reproduce. This checklist is for the final review of a completed brief. It tests whether a decision-maker can trace the comparison from question to evidence and understand what remains uncertain. It does not choose competitors or write the brief. It also does not decide whether a comparison is lawful in every jurisdiction. Obtain specialist review before using comparative claims publicly.
Prepare one review packet
Freeze the version under review and assemble the brief, source register, calculation sheet, captures or archived records, reviewer comments, and intended use. Record the decision the brief supports, its audience, the review date, the named owner, and whether any wording will appear in advertising or sales material.
Use one disposition for every check:
The U.S. Securities and Exchange Commission's business-description rule illustrates why competitive conditions and market demand can be material business information. UK comparative-advertising rules provide a more specific public-claims test: comparisons should be objective, material, relevant, verifiable, representative, and between products meeting the same need. These are source standards for review, not a conclusion that a particular brief or claim complies.
- Pass: the evidence supports the item as written.
- Repair: a specific correction or missing record is required.
- Escalate: authority, legal interpretation, or material uncertainty requires a named specialist.
- Not applicable: the reviewer records why the check does not apply.
Gate 1: decision and comparator set
Failure signal: a product appears only because its marketing language resembles the subject, or a premium edition is compared with a basic edition without disclosure.
- State the decision in one sentence. A brief for product positioning should not quietly become a procurement recommendation.
- Name every included competitor, product, edition, geography, and customer segment.
- Explain why each comparator belongs and record important exclusions.
- Confirm that compared offers meet the same need, or label the limits of the comparison.
- Check that the title, executive summary, tables, and recommendation use the same comparator set.
Gate 2: assertion provenance
Failure signal: a citation reaches a home page, search result, or undated summary that does not support the sentence.
- Give each factual assertion an evidence ID.
- Link to the primary source where available, with publisher, page title, publication or capture date, and relevant section.
- Preserve the quoted field, screenshot, or archived record needed to reproduce volatile facts.
- Separate source language from the analyst's interpretation.
- Label customer anecdotes, secondary reporting, and inference as such.
- Mark unsupported assertions for removal rather than hiding them behind a general bibliography.
Gate 3: like-for-like fields
Failure signal: one vendor's annual prepaid price is placed beside another vendor's monthly flexible price as if the units were identical.
- Define every comparison field before scoring it.
- Normalize currency, tax treatment, contract duration, usage unit, seat minimum, region, and effective date.
- Distinguish announced, beta, generally available, and discontinued capabilities.
- Compare product editions explicitly; do not borrow a feature from one tier and the price from another.
- Use “not found” only after documenting where and when the reviewer looked. Do not convert missing evidence into “does not exist.”
Gate 4: calculations and scoring
Failure signal: a “20% lower” claim uses the competitor price as the denominator in one section and the subject price in another.
- Put inputs, formulas, units, and rounding rules in the calculation sheet.
- Recalculate totals and percentages from source values.
- Check denominators, weights, missing-value treatment, and sign direction.
- Run a sensitivity check when a recommendation depends on an analyst-set weight.
- Confirm that qualitative labels follow a declared rubric and are not a disguised opinion.
- Reconcile every number in the executive summary to the table and workbook.
Gate 5: claims and contradictions
Failure signal: the summary says a competitor has no citation support while its documentation provides exportable references in a different edition.
- Match the strength of each claim to the evidence. Prefer “the reviewed documentation states” over an absolute claim when operation was not tested.
- Search for superlatives, exclusivity, causation, performance, security, and compliance wording.
- Check whether footnotes, limitations, or later sections contradict the summary.
- Record credible contrary evidence and explain its effect.
- Separate observed fact, calculation, interpretation, and recommendation.
- Route externally used comparative claims to the appropriate legal or communications reviewer.
Gate 6: completeness and usability
Answer the stated decision, not merely list features.
- Include the comparison date and a clear refresh trigger.
- Show evidence gaps and the consequence of each gap.
- State assumptions, excluded use cases, and residual uncertainty near the recommendation.
- Give the decision-maker a next action, owner, and deadline.
- Remove duplicated tables and decorative scoring that does not change the decision.
Gate 7: approval, release, and expiry
Record content reviewer, calculation reviewer, accountable owner, and any specialist approval.
- Confirm approval covers this exact version and intended audience.
- Restrict internal-only material from public or customer-facing reuse.
- Assign an expiry date based on volatility, not a generic annual cycle.
- Define refresh triggers such as price changes, edition changes, acquisitions, product launches, or contradictory evidence.
- Retain the final brief and review record together.
Worked artifact: catch four defects before release
A synthetic team reviews a brief comparing three research assistants for a renewal decision. The brief concludes that Cedar is “the only tool with traceable citations” and “18% less expensive.”
| Check | Evidence reviewed | Finding | Disposition | | --- | --- | --- | --- | | Comparator edition | Pricing captures and feature pages | Cedar Pro is compared with Alder Basic | Repair the edition table and recommendation | | Citation claim | Current product documentation | Alder Enterprise exports citations; the brief reviewed Basic | Remove “only” and state the edition boundary | | Price calculation | Workbook formulas | Annual prepaid Cedar price is compared with monthly Alder price; denominator also differs | Recalculate on annual equivalent terms | | Freshness | Captures and change log | Birch pricing capture is 11 months old and its page now redirects | Refresh before decision | | Contrary evidence | Source register | Analyst recorded a Birch limitation but omitted it from the summary | Add the limitation and explain relevance | | Release authority | Approval record | Internal strategy approval exists, but no approval covers sales use | Keep internal; escalate external wording |
The brief fails review even though its layout is complete. The owner corrects editions and units, replaces the exclusivity claim, refreshes Birch evidence, and obtains a second calculation review. The revised recommendation now says that Cedar has the strongest citation workflow among the editions reviewed as of the stated date, with an unresolved Birch evidence gap. That narrower statement is reproducible and honest about uncertainty.
Final sign-off record
Attach the completed checklist with the brief version, evidence register hash or version, open repairs, residual uncertainty, intended audience, approvers, approval dates, expiry, and refresh triggers. A pass means the defined checks succeeded for this version and purpose. It is not proof that every market fact is complete or that future claims remain current.
Two Prune can assess how a participant compares evidence, checks calculations, identifies contradictions, and communicates uncertainty in a work simulation. It does not verify an organisation's market claims, certify advertising compliance, or make employment decisions.
Sources
U.S. Securities and Exchange Commission, 17 CFR 229.101, Description of Business
- UK Government, Marketing and advertising: the law
- UK Business Protection from Misleading Marketing Regulations 2008, regulation 4
Sources: [1] [2] [3]
Sources
- 1.U.S. Securities and Exchange Commission, 17 CFR 229.101, Description of Business · Electronic Code of Federal Regulations
- 2.UK Government, Marketing and advertising: the law · UK Government
- 3.UK Business Protection from Misleading Marketing Regulations 2008, regulation 4 · www.legislation.gov.uk