Guide library

Protocol note 04

The claims-to-evidence ladder for marketing copy

Match the strength of a claim to the strength of its support.

ClaimsEvidenceCopy review

Messages become riskier one adjective at a time. “Helps organize” becomes “works faster,” then “the fastest,” even when the evidence never changes. The claims-to-evidence ladder makes that escalation visible.

Four levels of claim strength

Level one is a descriptive feature: “Exports a CSV” or “Includes three planning templates.” Verify the current product version, access conditions, actual output, and important exclusions.

Level two is a qualified functional benefit, such as “Helps organize message-test planning.” Define what the benefit means, for whom, under which use conditions, and what evidence connects the feature to that outcome.

Level three is quantified or comparative performance: “Reduces setup time by 20%,” “produces fewer errors than spreadsheets,” or “twice as fast.” Support must cover the stated metric, population, task, comparator, versions, and operating conditions.

Level four is absolute, universal, or best-in-class: “fastest,” “best,” “works every time,” or “eliminates errors.” Its scope is broad. Unless the evidence genuinely covers that scope, narrow or remove the claim. A level-one statement can still mislead when a feature is restricted, obsolete, or contradicted elsewhere; the ladder indicates breadth, not automatic acceptability.

Use a written claim-review instrument

Review the whole asset, including headline, image, demonstration, caption, call to action, and nearby qualifiers. Record the exact promise and the overall takeaway a reasonable reader could draw, not only the interpretation the copywriter intended.

For each material statement, record the product version, population, geography, use case, duration, metric, comparator, evidence owner, completion date, covered scope, contrary information, and next review trigger. Evidence should exist before the claim is disseminated; later support does not rewrite what was available at publication.

  • Exact claim: which words, numbers, visuals, or demonstrations create the promise?
  • Likely takeaway: what would a reasonable reader believe the product will do, for whom, and under what conditions?
  • Claim type: feature, benefit, quantified outcome, comparison, superiority, guarantee, or testimonial implication?
  • Method match: did the evidence measure the promised outcome under relevant conditions?
  • Coverage: does support match the number, comparator, duration, population, and product version?
  • Qualification: can a necessary boundary appear clearly beside the claim without contradicting it?
  • Decision: approve as written, narrow to the evidenced scope, hold pending evidence, or remove?

Match evidence to the promise

Support should correspond to the claim’s actual meaning, not an adjacent fact. Product quality assurance can verify that a CSV export works; it does not establish that users complete work faster. An internal preference poll does not establish improved accuracy. A comparison with an old spreadsheet process does not support “faster than any platform.”

For a performance test, document the protocol, product and comparator versions, target population, recruitment, task, outcome definition, observation period, exclusions, missing-data treatment, analysis, and uncertainty. If a survey supplies evidence, preserve the exact wording, sample source, mode, dates, achieved sample, weighting, sponsor, and limitations.

Ask what the published message means in participants’ own words before supplying interpretations. A persuasive message that reliably creates an unsupported takeaway is not successful, and an interpretation study does not substitute for evidence that the underlying performance claim is true.

Worked example: a broad claim with narrow support

Consider the hypothetical line: “Build better surveys twice as fast as any other platform.” Assume the record contains product checks confirming templates and CSV export, plus an internal timed exercise against the team’s previous spreadsheet process. No external platform was tested, and “better” was never defined.

The sentence makes at least four promises: the product supports survey building; output is better; completion is twice as fast; and it outperforms every competing platform. The universal comparison puts it at level four, while the available record verifies only features and one narrow internal workflow.

A more auditable line would be: “Plan a survey in one workspace with a questionnaire outline, disclosure checklist, and CSV export.” Each feature still needs verification on the advertised version. A later speed study would need a defined task, population, comparator, timing endpoint, exclusion rules, sample plan, and analysis before any actual result could be stated.

Failure modes that weaken the evidence record

Use the claim register to stop these patterns before publication:

  • Adding adjectives while reusing evidence collected for a narrower claim.
  • Publishing first and attempting to substantiate later.
  • Treating testimonials, customer letters, or preference ratings as performance tests.
  • Measuring satisfaction while claiming speed, accuracy, or revenue impact.
  • Using employees or expert users to support a claim aimed at ordinary customers.
  • Comparing with an obsolete or selectively weak alternative.
  • Selecting a favorable endpoint after reviewing results.
  • Omitting mixed findings, exclusions, attrition, or uncertainty.
  • Using a distant qualifier to contradict an unqualified headline.
  • Letting product, comparator, audience, or market changes silently age out the evidence.

Review triggers and limits

Evidence has a lifespan. Reopen a claim when the product, price, audience, comparator, method, or market changes. Assign an owner and a date or event that triggers review; do not rely on someone remembering why an old line was approved.

The ladder does not decide whether evidence is legally sufficient and does not replace category-specific review. Meaning depends on the full presentation. Canadian law expressly considers general impression as well as literal meaning, while U.S. substantiation expectations depend on factors including the claim, product, potential harm, cost of substantiation, benefits of a truthful claim, and relevant expert practice.

Message Test Bench has not used this framework to certify a live advertiser’s claims. It is an editorial planning tool, not legal advice or an evidence hierarchy mandated by a regulator.

Sources and scope

  1. FTC Policy Statement Regarding Advertising SubstantiationU.S. Federal Trade Commission

    Foundational policy that express and implied objective claims need a reasonable basis before dissemination.

  2. Competition Act, section 74.01Justice Laws Website, Government of Canada

    Canadian statutory text covering materially misleading representations and performance claims lacking an adequate and proper test.

  3. Performance Claims Not Based on an Adequate and Proper TestCompetition Bureau Canada

    Practical guidance on claim fit, prior testing, relevant controls, subjectivity, and real-world conditions.

Sources support the specific statements described above; they do not validate this publication’s rubric, guarantee a compliant execution, or replace context-specific professional advice.

What this page is: a research-methods guide, not a report of completed consumer fieldwork. Corrections and material revisions are recorded under the publication’s editorial standards.