Every claim, checked against the source.
Building a way to read every factual claim in live advertising against the specifications published on the advertiser's own website, so the copy and the product stay in step as a range changes.
Speak with NicBritish Consumer Manufacturer
Client name, products and figures withheld for confidentiality.
The problem on the table.
Advertising copy is written once and served for years. Product specifications change, environmental claims are updated as the evidence firms up, and the ads carry on saying what was true when somebody wrote them. Nobody owns the checking, because it sits between the people who write the ads and the people who maintain the website.
The reason to check is that an advertising claim is only as good as the evidence behind it. A performance or environmental claim that has drifted from the company's own published position is worth catching early, and it is exactly the kind of thing an advertiser would rather find themselves than have raised for them.
The account also carried years of switched-off advertising, and an audit that treats a paused ad as a finding buries the live faults underneath history nobody can act on.
What forensic looked like here.
Scope came first. Every asset was checked for whether it is serving now, and the audit was restricted to what a customer can currently be shown. A large majority of the ads in the account were switched off, so that single decision was the difference between a short list of real problems and a long list of archaeology.
Each serving claim was then read against the client's own live product pages, fetched the same morning, rather than against a specification sheet or a memory of what the product does. Where the two disagreed, the client's own published page was treated as the authority.
Findings were graded by what could be evidenced. A claim that contradicts the client's own page is confirmed. A claim the page cannot settle is reported as unverifiable rather than presented as a fault, because a pack that overstates its findings gets picked apart on the one row that was not solid.
Two candidate findings were withdrawn during the check and recorded as withdrawn, because a finding raised and quietly dropped is how a reader stops trusting the rest of the list.
- A serving-assets-only scope, so paused history never buries a live fault
- Every claim read against the client's own live product pages
- Findings graded confirmed, unverifiable or withdrawn
- Landing-page destination checks against what each ad promises
- A covering note pitched for the person who has to act on it
What changed.
The output is a short, evidenced list rather than a long one: each item traceable to the page that settles it, alongside a record of what was checked and found correct. Anything the pages cannot settle is reported as unverifiable rather than as a fault, so the list can be worked line by line.
The work also produced a repeatable method. Because the check reads the client's live pages rather than a static specification, it can be re-run whenever the product range moves, which is the point at which advertising copy usually stops being true.
- Serving onlypaused history excluded
- Live pagesthe source of truth
- 3 gradesconfirmed, unverifiable, withdrawn
- Re-runnableas the range changes
Other engagements.
Your number is next.
A proper conversation with the person who would do the work. Tell me where you are and what you are weighing up, and you will get an honest, forensic read on what I would change, what I would leave alone, and where the real upside is.
Speak with NicThe senior read,
once a week.
Most marketing newsletters are noise on a schedule. This is not that. Once a week I send a short, considered read on the shifts that actually change what you should do, across the AI shift, PPC, technical SEO and analytics. Filtered, weighted and signed off by me before it goes out. The judgement is mine, and now you can hear every edition in my own voice.
Nic CraneFounder, The Crane Consultancy
Join readers of The Crane Standard.