Before the headline: build an evidence record to validate digital PR angles
Learn how to document the evidence behind a digital PR idea before producing it, pitching it to the media, or proposing a collaboration.
A compelling idea is not always a publishable or verifiable story
In digital PR, it is easy to fall in love with a potential headline before checking whether the story can stand up. An interesting observation, a high figure or a surprising comparison may be a useful starting point for research, but it is not yet a conclusion ready to produce, pitch or distribute. A publishable angle needs more than appeal: it must state a specific claim, be supported by traceable evidence, and honestly explain what that evidence does and does not demonstrate. It also needs defensible editorial context. A topic being popular does not mean it is new, relevant to an audience or suitable for a particular publication. Validation does not mean predicting coverage, links, mentions or SEO results. Its role is more practical: to reduce the risk of investing hours in analysis, design, content, outreach or collaborations around a conclusion that cannot later be verified or that requires too many caveats to communicate rigorously.
- An initial idea is a working hypothesis, not proof.
- A headline must not be broader than the data supporting it.
- Validation helps you decide whether to proceed, investigate further, reframe or discard the idea.
The risk of mistaking an observation for a campaign conclusion
An observation may be valid and still not justify a broad claim. For example, identifying changes in a limited set of searches does not in itself demonstrate a national trend; finding a difference between two groups does not explain its cause; and an internal survey does not automatically represent the whole population. The problem usually emerges when information is lost as analysis is turned into a story. The period, metric definition, geographic scope, sample or exclusion criteria are omitted. The result may sound definitive, but it is open to reasonable questions: where does the data come from? Which population does it cover? When was it collected? What exactly was compared? Are there alternative explanations? An evidence record requires these answers to be retained before the idea reaches the creative stage. It is a quality-control measure, not a brake on creativity.
- Do not turn correlation into causation without a methodology that allows you to do so.
- Do not extrapolate local, sector-specific or sample-based findings to a larger population without stating this.
- Do not present estimates as direct measurements.
- Do not use a secondary source as the sole support when the original document is available.
The evidence record: a pre-production sheet for every digital PR angle
An evidence record is a review sheet that accompanies each angle from its initial formulation through to the decision on whether to produce it. It is not intended to replace legal, statistical or sector-specific review where that is needed. Its purpose is to make the chain between a claim and its evidence visible. It is useful to create one sheet per angle, even when several ideas originate from the same dataset. Each may have a different scope, relevance and level of robustness. Centralising this information makes reviews across SEO, content, PR, analytics and approval stakeholders easier, and prevents the rationale for an idea from being scattered across spreadsheets, messages and disconnected documents. The sheet should be readable without additional context: anyone reviewing it should understand what is being claimed, what supports it, under which conditions, and which risks remain.
- Angle identifier and version.
- Core claim and provisional headline, clearly separated.
- Sources, links, access date and review owner.
- Methodology, definitions and calculations applied.
- Limitations, risks, editorial novelty and the decision made.
Step 1: write the core claim with precision and limits
Start with a single declarative sentence, not a slogan. Include the subject, observed metric or behaviour, geographic scope, period, and the population or universe analysed where relevant. If any of these elements is not yet known, mark it as pending rather than concealing it. A constrained statement is often more useful than a grandiose one. “In the analysed sample, X increased between dates Y and Z” is verifiable if the sample and calculation are documented. “X has surged in Spain” requires a definition of “surged”, sufficient coverage and a clear comparison baseline. Then separate the core claim from the provisional headline. The headline can be creative, but it must never introduce causality, universality or certainty that is not present in the validated claim.
- Avoid causal verbs such as “causes”, “proves” or “explains” if you only have an association.
- Define ambiguous terms: “affordable”, “common”, “better”, “record” or “significant growth”.
- Specify the reference period and comparison point.
- Note what would need to happen for the claim to be considered false or insufficient.
Step 2: link every part of the claim to its evidence
Break the claim down into verifiable components. If you say something has changed, you need a time series or two comparable measurements. If you compare territories, you need equivalent data for both. If you refer to people, you need to know how the population was defined and selected. If you use your own calculation, document the formula, filters and decisions that affect the result. For each component, record the primary source or the source closest to the origin, the URL or document location, publication date, access date and any conditions of use. Where possible, retain a working copy or stable internal reference: pages are updated, dashboards change and some data becomes inaccessible. A reputable source does not remove the need to check whether it answers the exact question. The authority of the organisation and methodological suitability are separate criteria.
- Which exact part of the claim the source supports.
- The original data, table, methodology or statement that must be reviewed.
- The data date versus the publication date.
- Transformations made: percentages, averages, rounding, groupings or exclusions.
- Evidence confidence level: direct, partial or insufficient.
Steps 3 and 4: classify the data and review its scope before interpreting it
Not all evidence has the same nature. Labelling the type of evidence prevents an interpretation from accidentally gaining the status of data. A simple classification distinguishes between first-party data, external sources, estimates and interpretations. First-party data comes from an organisation or research carried out for the campaign. It should be accompanied by its collection method, universe, period and limitations. External sources may include public statistics, studies, databases or specialist publications; review their original methodology. Estimates depend on assumptions and should be presented as such. Interpretations are reasoned readings of findings, not standalone facts. Next, check recency, geographic scope, sample size and composition, variable definitions and possible bias. Also ask what is left out of the measurement. A figure may be accurate for its own universe and still not be useful for the headline you want to support.
- Is the data still current for the proposed story?
- Does the geographic scope match that of the claim?
- Is the sample sufficient, and how was it selected?
- Are equivalent periods, groups and methods being compared?
- Are there definition changes, seasonality, self-selection or missing data?
- Does the source state margins of error, uncertainty or relevant restrictions?
Steps 5 and 6: record the nuances and assess editorial novelty
Nuance does not necessarily weaken a proposal; it often makes it more credible and more interesting. Record exceptions, alternative explanations, subgroups that do not follow the pattern and claims that must be avoided. If a limitation is essential to understanding the result, it should be part of the angle rather than hidden in a final note. Then assess editorial novelty. Do not ask only, “Is it trending?” Ask, “What does this story add that an editor, audience or reader did not already know?” Novelty may come from updated data, an underexplored comparison, transparent methodology, a practical consequence or a specific geographic perspective. It may also be absent, even if the figure is correct. Review recent coverage of the issue, which claims are repeated, and whether your evidence genuinely changes the conversation. This review cannot predict editorial interest, but it helps avoid derivative proposals or proposals framed as though they reveal something already widely known.
- The nuance required to communicate the story accurately.
- Alternative explanations that cannot be ruled out.
- Likely questions from journalists, editors or experts.
- What is genuinely new, and for whom it is relevant.
- Risk of repetition, seasonality or topic saturation.
Frequently asked questions
What is an evidence record in digital PR?+
It is a sheet that documents an angle’s claim, the sources supporting it, methodology, scope, limitations, risks and the decision to proceed, reframe or discard it. It enables the proposal to be reviewed and defended using shared criteria.
How do I decide whether to reframe, investigate further or discard an angle?+
Reframe it when the evidence supports a narrower, more precise claim. Conduct further research when a key missing data point can be obtained through a viable and proportionate method. Discard it when the claim relies on excessive assumptions, sources are not traceable, bias is decisive or there is no defensible editorial novelty.
What should an evidence record template include?+
At a minimum: the core claim, provisional headline, objective and audience, sources and dates, evidence type, methodology and calculations, geographic and time coverage, sample or universe, limitations, permitted and non-permitted interpretations, editorial novelty review, owner and final decision.
How should I present a validated angle without promising results?+
Explain what the research claims, how it was conducted, its scope and its limitations. You can state that the material is prepared for rigorous editorial communication, but do not promise coverage, mentions, links, traffic, authority or ranking improvements.
What changes if the activity is a sponsored collaboration?+
A paid collaboration requires a different approach from an earned editorial story. It should be communicated transparently so that the audience and publication understand its commercial nature. Depending on the case, links may need rel="sponsored" or rel="nofollow" attributes, in line with the applicable Google Search Central guidance on classifying outbound links.
Sources and references
- Google Search Essentials — Google Search Central
- Spam policies for Google web search — Google Search Central
- Qualify outbound links — Google Search Central