A Survey Doesn’t Become News Just Because It Has Percentages
A practical guide to designing surveys that produce publishable findings, with a well-defined sample, carefully worded questions and transparent methodology.
Percentages alone are not enough to support a news story
A survey can make a good digital PR asset, but only if its data answer a relevant question and can be interpreted rigorously. A striking figure does not, by itself, prove that a finding represents a population, is novel or implies causation. Before commissioning the work, consider what claim you would want to make and what evidence you would need to support it.
- Formulate a specific, useful question for the audience you want to reach.
- Write down in advance which conclusions would be reasonable and which would fall outside the study’s scope.
- Assess the editorial interest of the question, not just the possibility of producing a striking percentage.
Define whom the survey represents
The target population is the group you want to discuss: for example, adults living in a particular area or professionals in a sector who meet defined criteria. You then need to explain how respondents were selected. A probability sample, in which selection allows the chances of inclusion to be known, is not the same as a panel, an open invitation or a survey shared with followers. The latter can be useful for exploring opinions, but should not automatically be presented as representative of an entire population.
- Specify the relevant location, time period and inclusion criteria.
- Describe the recruitment channel, who could take part and whether incentives were offered.
- State the number of valid responses and how incomplete or excluded responses were handled.
- If you apply weighting, explain which variables it uses and for what purpose; weighting does not necessarily correct every bias.
Look for bias in the sample and questionnaire
Bias can arise before anyone answers: some people may have more opportunities to receive the invitation or a greater interest in taking part. It can also result from non-response, the order of questions or wording that suggests an expected answer. A large sample does not eliminate these problems. Reviewing the entire process helps distinguish a pattern in the survey from a pattern that may exist beyond it.
- Check whether the recruitment channel excludes groups that matter to the question.
- Consider who did not respond and whether their absence could affect how the results are interpreted.
- Avoid double-barrelled questions, ambiguous terms, and answer options that are incomplete or overlap.
- Check whether the context or order of questions could influence responses.
Test the questionnaire before collecting responses
Pre-testing can reveal problems that are easy to miss when the team simply reads the questionnaire aloud. Ask people similar to the target population to complete the survey and explain what they understand by each question, how they choose an answer and whether any options are missing. If you make substantial changes to the questionnaire after fieldwork begins, document the change and consider whether responses are still comparable.
- Check that instructions and concepts are understood as intended.
- Measure completion time and identify questions that cause people to drop out or become confused.
- Review skip logic, “don’t know” or “not applicable” options, and the mobile experience.
- Keep a dated version of the questionnaire used.
Analyse results without going beyond what the data say
Start with the overall result, then check how the subgroups are formed. If a category contains few responses, an apparent difference may be unstable and may not justify a headline. Nor should you turn an association into a causal explanation: observing two responses together does not prove that one caused the other. If you report statistical uncertainty, make sure the method and design allow it to be calculated; for non-probability samples, the usual margin of error for a random sample should not be applied automatically.
- State the response base for each percentage, especially for cross-tabulations and subgroups.
- Distinguish exploratory results from confirmatory conclusions.
- Avoid selecting only the cross-tabulation that produces the most striking figure without explaining that multiple comparisons were explored.
- Check that rounding, denominators and charts match the original data.
Publish the methodology, limitations and uncertainty
Every finding needs context so that someone else can understand what it measures and how far it can be generalised. Methodology is not a secondary note: it is part of the claim. Include who commissioned the study, who conducted it, when responses were collected, how the sample was recruited and what relevant limitations it has. If a figure comes only from participants in an open survey, say so clearly and avoid describing it as the opinion of the entire population.
- Present the finding alongside its target population, sample size and collection period.
- Explain the survey mode, recruitment, weighting and exclusions, where applicable.
- Describe specific limitations, such as a convenience sample or insufficient coverage of certain groups.
- Present intervals or measures of uncertainty only when they are appropriate to the design and have been calculated correctly.
Prepare materials that can be checked and cited
Make editorial work easier without turning the data into a claim that cannot be verified. A clear summary, a results table and a methodology sheet help reduce ambiguity. Include the exact question and answer options associated with each finding: a paraphrase can change the meaning. Keep a contact available who can explain the method, but do not suggest that an editor should accept a prewritten headline.
- Provide the exact question, calculation base and result you want cited.
- Make downloadable tables or aggregated data available where feasible and safe.
- Clearly separate the observed result from the brand’s interpretation or recommendation.
- Keep dated versions of the questionnaire, methodology and materials distributed.
Frequently asked questions
What makes a survey useful for digital PR?+
It answers a relevant question using an appropriate design, a verifiable result and clear methodological context. Editorial relevance also depends on the topic, the timing and each publication’s judgement; percentages alone do not guarantee coverage.
Is a large sample necessarily representative?+
No. Sample size alone does not correct selection bias, non-response or inadequate coverage of the target population. You need to consider how people were recruited and what conclusions that method supports.
Can I talk about a margin of error for any survey?+
Not automatically. Standard margin-of-error formulas require sampling assumptions that may not hold, for example with a convenience sample or an open survey. If the design does not justify calculating one, explain the limitation instead of giving a misleading figure.
Does an in-house survey have to be presented as sponsored content?+
Not necessarily. A study commissioned by a brand can be proposed as an editorial resource, but that does not make publication guaranteed coverage or a paid collaboration. If distribution involves compensation, its nature should be clear and it is worth checking the applicable labelling requirements. Google recognises attributes such as sponsored for advertising or sponsored links, and nofollow in certain cases.
Which sources should I consult when planning distribution?+
For search practices and link qualification, consult the Google Search Essentials documentation, Google Search spam policies and the Qualify outbound links guide: developers.google.com/search/docs/essentials, developers.google.com/search/docs/essentials/spam-policies and developers.google.com/search/docs/crawling-indexing/qualify-outbound-links. These guides do not replace editorial judgement or the obligations that apply to each collaboration.
Sources and references
- Google Search Essentials — Google Search Central
- Spam policies for Google web search — Google Search Central
- Qualify outbound links — Google Search Central