Editorial review method that separates supplier claims from checked evidence

This method makes the evidence trail visible before it produces a score. The ParentingInfluence.co.uk editorial research desk prepared it from public records on 5 September 2026. No item, account or service was tested. A named human reviewer must replace the desk credit before publication.

What to take away

Frame a testable question

Name the exact reader, product or service, intended use and conditions. 'Which travel cot is easiest?' is too loose. A usable question specifies the model, journey, person carrying it, storage limit and the ordinary setup task. For an agency, define whether the reader needs representation, a single campaign or creator sourcing.

List the exclusions beside the question. Do not generalise an adult-only desk check to use by a child, a disabled parent or a family with a different safety need.

Keep evidence in separate columns

Use four labels: supplier statement, official or independent record, direct observation and editorial judgement. A first-party service page can establish what the provider says it offers. Mumsnet's current Influencer Network page, for example, describes a UK family-campaign network and brand-matching service. It does not prove campaign quality or suitability for one creator.

Search public records that match the subject. The OPSS product alert and recall service can reveal a specific UK safety record. Companies House can confirm a registered entity and filings. Neither search supplies a positive performance result when no adverse item appears.

Publish the test protocol first

Define the sample source, model or account, settings, test place, dates, repetitions, measuring instrument and stop rule. Start a digital trial with fictional names and adult-only media. A later family test needs a written purpose, minimum data, access control, withdrawal route and specialist approval.

For every result, preserve the raw observation. A sentence such as 'setup took four minutes' needs the start point, end point, attempts and conditions. If nobody timed the task, report the documented steps instead of inventing a duration.

Add a deviation log while the work is under way. If the reviewer changes a setting, replaces a damaged sample or abandons a repetition, record the event before continuing. The editor can then decide whether the observation remains comparable or needs a fresh run. Never delete an inconvenient attempt simply because the final trial worked.

The ICO says its Children's code design tests are guides and not official assessments. A completed questionnaire must not be presented as certification.

Score only comparable evidence

Set mandatory gates before weighted criteria. Safety identity, permitted use, rights, privacy and data export may be pass conditions. A failed gate should remain visible rather than disappear inside an average.

For preferences, publish the scale and anchors. 'Five' might mean every required file exported with names and metadata intact; 'three' might mean the main files exported but manual reconstruction was needed. Weight the criteria for the declared use, then show the observations as well as the total.

Control commercial influence

Record who supplied the item, the normal purchase route, payment, loan period, affiliate commission and editorial approval rights. The ASA's current testimonials and endorsements advice requires genuine records and warns against omitted commercial interests. Objective claims still need evidence when they appear inside a testimonial.

Before release, a second editor should trace two conclusions to raw evidence and challenge the verdict. Publish what was not examined, the date and the event that will trigger a recheck. The review stays on hold if the sample identity, conflict record or decisive test cannot be reconstructed.

Before you act

Common questions

What must a reviewer do before publication instead of the desk credit?

The article says a named human reviewer must replace the desk credit before publication. The desk prepared the method from public records on 5 September 2026, and no item, account or service was tested. So the desk credit alone is not enough for a published review.

Why can a supplier page not prove campaign quality?

A first-party service page can establish what the provider says it offers. The article cites Mumsnet's Influencer Network page, which describes a UK family-campaign network and brand-matching service. It does not prove campaign quality or suitability for one creator, so the claim stays in the supplier column.

What happens if a decisive test cannot be reconstructed?

The review stays on hold if the sample identity, conflict record or decisive test cannot be reconstructed. Before release, a second editor should trace two conclusions to raw evidence and challenge the verdict. The article also says to publish what was not examined and the date for a recheck.