Operations
Part of How do you run digital PR operations that survive client scrutiny?
Check the data and sources in a digital PR launch review
Compare vendor claims with independent evidence before you sign off a digital PR launch review, and see what an assessment can and cannot tell you.
What to take away
- Decide first what the review is forto approve a launch, pause it, or re-scope the budget.
- Ask for the assessed date range, the sources used, and the reviewer's name.
- Keep vendor claims and independent evidence in separate columns.
- Expect a limits section covering what was not tested and what data was unavailable.
- Sign off only when every claim maps to a source you can open.
Set the scope before anyone writes the report
A launch review answers one narrow question: do the data and sources behind this campaign hold up? Write that question down before the reviewer starts, or the report drifts into brand opinion.
Four things to agree at kickoff
- Dates under assessment
- Channels in scope
- Sources allowed
- Owner of final decision
Compliance belongs in scope too. For non-broadcast marketing, check the CAP Code and relevant ASA guidance; where a campaign involves personal data, check the UK GDPR, the Data Protection Act 2018 and ICO guidance. The ASA's work on monitoring ads and acting on breaches means a review that never opens the ad copy is incomplete.
What the review can and cannot assess
This article assesses a process, not a product. It looks at how a launch review should be scoped, sourced and signed off. No vendor was tested, no platform was trialled, and no analytics account was opened.
An editorial review can check published material: press releases, landing pages, dataset notes and the sources they cite. It can confirm whether a figure in a pitch matches the study it came from, such as the ONS Personal well-being in the UK bulletin and its accompanying datasets.
For a worked check, take a draft claim that a region has the highest average life satisfaction. Find the matching measure in the ONS Personal well-being in the UK data, confirm the geography and period, then reproduce the ranking using the same measure; record the dataset, table and source date, and do not sign off if the result cannot be reproduced. For a claim about a company's status or incorporation, check the Companies House register rather than relying on a press release repeating it.
It cannot inspect private analytics, CRM records or a measurement provider's internal method unless the reviewer is given access and the scope permits it. For campaign outcomes, Google Analytics 4 and Google Search Console can provide web-analytics and search-performance data, while Meltwater or Cision can provide media-monitoring records; record the export date and method, and do not treat tool output as independent proof. Our digital PR operations and delivery guide for 2027 shows where review sits inside the wider delivery cycle, which helps if you are rebuilding the workflow around it.
Seven steps to run a launch review
The sequence below keeps the review auditable. Each step should produce a file someone else can open.
Seven steps to run a launch review
- Fix decision date and sign-off owner
- List every public claim word for word
- Trace each claim to a primary source
- Check definitions before reusing audience lists
- Recalculate one figure by hand
- Record gapsno source, no date
- Publish review with limits and initials
- Fix the decision date and name the person who signs off.
- List every claim the campaign makes in public, word for word.
- Trace each claim to a primary source, such as the ONS Personal well-being in the UK bulletin and accompanying datasets; record the source title, publisher, table and publication date.
- Where a claim involves personal data, check the relevant definitions and handling against the UK GDPR, the Data Protection Act 2018 and ICO guidance.
- Recalculate one figure by hand from the raw table; for survey-based claims, record the stated sample size and denominator, or mark them unavailable.
- Record the gapsclaims without a source, sources without a date, and dates that moved.
- Publish the review with a limits section and the reviewer's initials.
Decide how to label each claim
Mark every claim as measured, modelled or asserted. Measured claims need a method note. Modelled claims need stated assumptions, and asserted claims stay in the commentary.
Agree what counts as a gap
A gap is a claim you cannot recheck, not a claim you dislike. Write the definition down before the review starts, so the findings do not depend on the reviewer's taste.
Compare figures against a stated method
Audience numbers cause most of the argument in a launch review. A placement gets reported as reaching a large audience, with no note on panel size or collection method.
Nielsen's documentation of digital content ratings sets out how digital audiences are measured. Use that kind of source to ask sharper questions, not to settle the number.
Two questions do most of the work. Was the figure supplied by the platform, or estimated independently? If the platform supplied it, label it a vendor claim and say so in the report.
Where to record the outcome
Keep the register in a shared Google Sheets or Microsoft Excel workbook the delivery team can access. Record the source date and reviewer initials alongside each claim; a claim with no source date cannot be rechecked next quarter, which matters when a client asks for a refresh.
Our quality checklist for data and sources gives you column headings to reuse, so the register and the checklist stay aligned.
Use these register and checklist column headings: claim wording; claim label; source title and URL; publisher; dataset or table; source publication date; data period; definition and geography; method and assumptions; sample size or denominator, if stated; recalculation; vendor-supplied status; gap or limitation; owner; reviewer initials; sign-off decision and date.
Assign ownership for each gap. One person closes it and another rechecks it. Without that split, gaps stay open.
Common questions
Does a launch review need a named reviewer?
Yes. Someone must own the method, the source list and the limits. Without a name attached, findings get renegotiated in the next meeting.
Can a review run after the launch?
It can, though gaps cost more to close once coverage has published. A post-launch review is still worth running if the campaign will repeat.
How many claims should a review sample?
Cover every channel and every headline figure, and record how many claims were checked out of the total in scope. State the selection rule and check any individual claim that the sign-off decision depends on.
What should a limits section say?
It should name what was not tested, which data was unavailable, and which claims rest on vendor material.
