Blog

What to check before you pay for a PR survey

Decide whether a PR survey fits your goal, which one question it should answer, and how to tell if its findings are solid enough to publish.

By Andre Guelmann · Updated · 14 min read

Contents

In The B2B Numbers Census, a website-content audit of 614 B2B companies observed in September 2026 and published in October 2026 by ClaimTheStory Research, 7 in 10 companies used someone else’s numbers. Commissioning a PR survey is one way to generate numbers of your own, yet an expensive questionnaire with weak screening, vague answer rules or tiny group bases may not survive a journalist’s questions.

Before you sign a supplier’s proposal, you need to know whether a PR survey is the right route, which single question to commission it to answer, what to ask the supplier upfront, and how to judge the delivered tables before publishing. This guide walks through that process in six sections, in order: what a PR survey can and cannot show, a four-part commissioning test, defining one core question, seven questions to ask a supplier, checking the delivered results, and setting your publication policy and methods note.

What a PR survey is, and what it can and cannot give you

A PR survey is a survey commissioned so its findings can be published to journalists and other outside audiences, usually fielded by a research supplier on an online panel. The same survey can support both press outreach and content marketing; the difference on this page is your decision as the buyer commissioning and evaluating the work, rather than designing and running the survey yourself (covered in survey research for content).

Survey answers show what a defined group of respondents reports thinking or doing, with one limit: answers aren’t checked against operational or financial records. A survey cannot prove causes or measure exact operational counts from logs. Whether the results describe a wider population depends on how the sample was drawn: probability-based online panels exist, while the example on this page uses an opt-in (non-probability) panel whose findings describe only the clinics that answered.

What a PR survey can and cannot give you

Can give you

What a defined group of respondents reports thinking or doing, in their own answers.

Cannot give you

  • Proof of causes.
  • Exact operational counts that only records hold.
  • A picture of a wider population, when the panel is opt-in.

Is a PR survey worth commissioning? A four-part test

Before asking suppliers for quotes, run your plan through four tests in order:

  1. Audience decision: Does a specific decision that your readers or buyers make depend on the answer?
  2. Existing records: Can data your company already holds, or public records, answer the question instead (see how to get data for content)?
  3. Respondent knowledge: Can the people you plan to invite actually know the answer?
  4. Budget for group comparisons: Can your company afford a sample large enough to report the subgroup comparisons you plan to publish?

Original research can give you numbers of your own, but it costs supplier fees, internal review time and opportunity cost, and no survey can guarantee media coverage, citations or a financial return. For a full look at weighing costs and benefits before investing, see original research for content marketing.

What to ask the survey to investigate: one question, not a headline

Commission a supplier to investigate one specific question, not to prove a predetermined headline. Everything PR’s PR survey guide advises: “Build the survey backward from one headline and one core question.” Drafting a possible headline before fieldwork is useful to test whether an answer would matter to journalists (covered in data-led PR), but the supplier is briefed on the neutral question, the question is never worded to nudge respondents toward a hoped-for result, and the outcome stays unknown until fieldwork closes.

If you are still choosing between several possible questions, use B2B research ideas before writing a supplier brief.

The commissioned question, part by part
  • Who answers An owner or manager overseeing new-patient scheduling.
  • Who acted The clinic.
  • What was turned away A new patient referral.
  • When the 14 days start On receiving the referral.
  • Period The 30 days before answering.
  • Answer rule Yes if aware of one; No only if able to confirm none; otherwise Not sure.

What to ask a supplier before you sign

Put seven questions to any survey supplier before signing a proposal:

  1. Sample source and recruitment: Who are the respondents, where do they come from, and how is the panel recruited (probability-based or opt-in)?
  2. Screening: What opening questions filter respondents, and how is one response per organization enforced?
  3. Group counts: How many completed responses are expected in each comparison group you plan to publish?
  4. Weighting: Will the results be weighted, and if so, against what reference figures?
  5. Wording and rules: Will you review and approve the full question wording, answer options and answer rules before launch?
  6. Deliverables: Do you receive the cleaned respondent-level data file, or only summary tables?
  7. Cleaning record: Will the supplier deliver a full cleaning log showing counts screened out, incomplete, duplicates and other quality removals, reconciled to the final retained responses?

How a panel is recruited governs what your findings can describe: an opt-in panel is not a random sample, so a conventional margin of sampling error does not apply to it. Ask the supplier how it documents recruitment and precision in its methods note. The American Association for Public Opinion Research (AAPOR) publishes best practices for survey research, covering design, data collection and analysis.

How the panel is recruited decides what the findings describe

Probability-based panel

Members are drawn at random, so a margin of sampling error can apply, with its own caveats.

Opt-in panel

People who signed up to take surveys. Findings describe the respondents only, and no margin of sampling error applies.

There is no universal minimum sample size for a PR survey. Where a supplier quotes a general floor (for example, Talker Research’s PR surveys page states “In order to have statistically significant data, you need a minimum of 80 respondents”), treat that as one supplier’s rule of thumb rather than a universal standard. The sample you need depends on the smallest subgroup you intend to publish and how finely you compare groups. For questionnaire design details, see survey research for content.

Judge the findings before you publish them

When the supplier delivers the tables, inspect the raw counts and cleaning log before accepting a single percentage.

The cleaning record, reconciled
  • Entered screening 486.
  • Screened out 104, leaving 382.
  • Incomplete 31, leaving 351.
  • Duplicate clinics 9, leaving 342.
  • Other quality removals 12, leaving 330 retained responses.
What the 42.6% measures

Measures

The share of clinics reporting at least one turned-away referral in the 30 days before answering.

Not measured

  • How often clinics turn referrals away.
  • How many patients were turned away in total.
  • How fast clinics book, or how long patients waited.

Once the findings pass these checks, see data-led PR to test whether the verified numbers support a press-facing angle.

Decide what to publish and what the methods note must show

Separate your internal working table from your public table before releasing anything. Under the team’s illustrative publication policy, the three qualifying size groups and their combined total are published with shares and bases, the respondent-level data file stays private, and the methods note may state that 24 responses came from clinics with 10 or more therapists while never revealing that group’s yes count, base or share.

Horizontal bar chart of the share of responding clinics that reported turning away at least one new patient referral in the past 30 days, by number of therapists: 1 therapist (base 58) 31.0%, 2–4 therapists (base 141) 41.8%, 5–9 therapists (base 92) 51.1%, total, 1–9 therapists (base 291) 42.6%. 10+ therapists: not reported: below the 50-answer threshold.
Larger clinics were more likely to report turning away a new patient referral. Opt-in panel, so the figures describe these clinics, not all US clinics. Illustrative data.

The full internal total (138 yes answers) is never published: the three included rows add up to 124 (18 + 59 + 47), so anyone who saw both could take 124 from 138 and recover the hidden group’s 14 yes answers.

How the internal total would leak the hidden group
  • Internal total 138 yes answers. Never published.
  • Published rows 18 + 59 + 47 = 124 yes answers.
  • The difference 14: exactly the 10+ therapists group’s yes answers.

Always assess disclosure risk with whoever owns the data and with the survey supplier before publishing; a minimum group threshold alone never makes survey records anonymous. Attach a complete methods note covering 14 items so journalists and analysts can inspect the work:

  1. Survey supplier and commissioning company
  2. Sample source and how the panel is recruited
  3. Fieldwork dates
  4. Reference period asked about
  5. Screening criteria and duplicate-removal rule
  6. Cleaning counts reconciled to retained responses
  7. Exact question wording and answer options
  8. Answer rule shown to respondents
  9. Bases for every reported group and total
  10. How not-sure and skipped answers are treated
  11. Group-size threshold and suppression rule
  12. Weighting (or confirmation that results are unweighted)
  13. Limits of the panel, self-report and measure
  14. Citation line and data-access policy

For how to structure the finished report and write that methods note in full, see writing up survey and research findings.


If you know what you want to say, we find the data that would support it, and tell you if somebody has already published it. Book a demo