Create one row for every material promise on the destination behind a ChatGPT ad. Record the exact wording, where it appears, what a customer could reasonably understand, the supporting record and the decision to keep, revise or remove it.
This is a working register for the destination page. The claim evidence matrix for ad copy covers the headline, body and call to action inside the ad. Filling in either register does not secure platform approval. The destination register’s purpose is to stop attractive wording from reaching the page without anyone knowing what supports it.
Capture the page as a customer encounters it
Review headings, body copy, buttons, captions, comparison tables, badges and images. A promise can be implied by their combination. A certificate badge beside every product might suggest wider coverage than its actual scope.
Include repeated claims in different locations. The first-screen statement, pricing section and mobile action panel may use slightly different wording. Treat them as separate occurrences linked to the same underlying claim so an edit does not leave a contradictory version elsewhere.
Save the page version or review date. Without that reference, a later editor may assume the approval applies to wording introduced after the review. The register should identify what was assessed, not merely record that “the landing page was checked.”
Translate the message into a testable proposition
Write what the statement asks the visitor to believe. “Ready when you are” might be harmless conversational copy in one setting and imply immediate availability beside a booking button in another. Consider the surrounding offer.
Distinguish factual statements from clearly subjective descriptions, then examine whether the subjective wording implies a fact. “Our best package” can be an internal preference, while “the best-performing package” suggests a comparison that needs a defined basis.
Keep the proposition close to the customer’s interpretation. Do not narrow it internally to an obscure meaning while leaving a broad headline on the page. The register should reveal that gap and lead to clearer copy.
Match the record to the advertised service
For each factual claim, identify the evidence owner and the record that supports it. A staffing schedule can support stated contact hours. A current specification can support a product attribute. A documented project may support a bounded account of that project.
Check product version, geography, period, customer type and operating conditions where relevant. Evidence for one configuration does not automatically cover every variant on the destination. Evidence from a former service process may not describe the current offer.
ICC’s substantiation article provides a reference for supporting factual claims. The detailed register structure here is an editorial method for applying that discipline to a specific page.
Resolve three different review outcomes
Use a hypothetical service page to rehearse the decisions. A weekday-support statement matches the actual staffed schedule and relevant agreement. It can be retained with clear hours and scope, provided the displayed wording stays within that record.
A “fastest in the industry” headline has no defined comparison set or suitable supporting study. The right decision is removal or replacement with a supported narrower statement. A colleague’s confidence that the service is quick does not fill the evidence gap.
A delivery statement has a documented time basis but omits that the period starts after the customer approves the artwork. Revise it to include the start point and applicable products. The claim may be usable after clarification even though its original presentation was too broad.
These examples are hypothetical review decisions. They are not claims about the site’s service, actual support availability or a measured delivery performance.
Three decisions in a hypothetical claim register
- Weekday support
Compare wording with the staffed schedule and applicable agreement.
Checkpoint - Fastest in the industry
Without a relevant comparison, remove the claim.
Reconsider - Delivery after approval
State the start point and which products the evidence covers.
Give each unresolved row a real next action
An unresolved claim needs more than a yellow status. Identify what is missing, who can supply it and whether the claim must remain off the destination until reviewed. A row that says “check later” can quietly become permanent.
Do not make the copywriter responsible for proving technical performance they cannot assess. The appropriate product, delivery or research owner should evaluate the record. The editor’s job is then to align the wording with the assessed scope.
Where a credential is involved, use the credential-context review to verify what it actually covers. Where the problem is placement rather than factual basis, use evidence beside the claim to keep necessary context visible.
Link the decision to the published wording
Record the final approved sentence or description, not just the draft that started the review. Include any essential qualification and the location of the supporting explanation. A small wording change can broaden the promise again.
Review the Swedish and English versions together. They can use natural phrasing while preserving the same factual scope. A cautious estimate in one language should not become a guaranteed outcome in the other.
Keep the register connected to the actual page version used by the campaign. If several destinations share a component, identify which ones contain the claim. That prevents an approved change in one place from leaving an unsupported promise active elsewhere.
Hand the register into maintenance
Before the page is used, every material row should have a clear disposition. Retained claims have an identifiable basis, revised claims match their final wording and removed claims are absent from the live candidate.
Then define what change would trigger another review: a new product version, changed staffing, expired credential or revised delivery process. The proof-maintenance guide covers that continuing responsibility.
The register is useful when another colleague can follow the chain from visible claim to evidence and editorial decision without relying on memory. It turns the destination’s promises into reviewable statements while leaving unsupported wording out of the customer’s path.
Sources and scope
Inventory and assess explicit and implied claims on a ChatGPT ad destination using exact wording, supporting evidence, scope and a decision before use.
Working methods and examples are editorial suggestions. Check current platform requirements and available features before implementation.