The Sentence Before the Statistic: A UK Buyer’s Wording Audit for Board-Bound Surveys

Boards seldom see the questionnaire. They see a percentage, a chart and a recommended course of action. That distance is useful for brevity but dangerous for governance: the wording that produced the number sits several steps upstream of the decision, often in an appendix nobody opens. For UK research buyers, the questionnaire is not a technical detail to be left entirely to the supplier. It is part of the evidence chain, and it deserves a procurement control before the board pack is locked.

Why wording belongs in procurement

A research commission is a purchase of evidence, not merely a purchase of fieldwork. The buyer specifies the decision, the audience and the timing; the supplier designs the instrument. Yet the instrument can quietly determine the answer. A question that frames a proposition positively, combines two ideas, leaves a term undefined or places one prompt before another can shift a headline result without any change in the market itself.

That is why the final questionnaire should be treated as a controlled document. It should be versioned, reviewed and retained alongside the data. If a board later asks why a satisfaction score moved, the first question should be whether the market moved or the sentence did. The exact size of any wording effect is an empirical matter, not a universal constant, but the risk is predictable enough to warrant a routine audit.

Four faults that survive the methods appendix

Leading framing

Leading questions do not always announce themselves. A phrase such as ‘our excellent service’ or ‘the convenient new process’ invites agreement before the respondent has made a judgement. A more neutral formulation separates the evaluation from the prompt.

Before: ‘How much do you agree that our new service is easy to use?’

After: ‘How easy or difficult did you find the new service to use?’

The second version still allows a positive answer, but it does not supply the adjective. For a UK consumer survey, the difference can matter when the board is deciding whether to extend a trial. For a B2B study, it can matter when an account team is defending a product roadmap.

Double-barrelled questions

A question that asks two things at once produces an answer that cannot be interpreted cleanly. ‘How satisfied are you with the speed and reliability of the service?’ conflates two performance dimensions. A respondent may be content with speed and frustrated by reliability, yet the single scale forces a compromise.

Before: ‘How satisfied are you with the speed and reliability of the service?’

After: ‘How satisfied or dissatisfied are you with the speed of the service?’ followed by ‘How satisfied or dissatisfied are you with the reliability of the service?’

The longer instrument is not automatically better, but it is more defensible. If board attention is focused on reliability, the buyer should know whether the headline figure is carrying unrelated satisfaction with speed.

Ambiguous terms

Words such as ‘recently’, ‘regularly’, ‘affordable’ and ‘decision-maker’ are common in UK surveys and equally common sources of confusion. One respondent’s ‘recently’ may mean the past month; another’s may mean the past three years. ‘Affordable’ may refer to list price, total cost or budget fit. ‘Decision-maker’ may mean the person who signs the contract, the person who recommends the supplier or the person who can stop the purchase.

Before: ‘Have you recently used a financial adviser?’

After: ‘In the past 12 months, have you used a financial adviser for advice about investments, pensions or retirement planning?’

The revised question is longer, but it gives the board a defined base. In B2B research, a role grid is often better than a single yes-or-no question: ask whether the respondent initiates, recommends, approves, vetoes or simply uses the category. That produces a buying-committee picture rather than a flattering count of ‘decision-makers’.

Order effects

Question order is part of questionnaire design, not an administrative afterthought. A prior question about a brand can prime later evaluation. A scale that runs from positive to negative can produce different responses from one that runs negative to positive. A long list of suppliers can advantage the first few names unless rotation is used.

Before: a brand-awareness list shown in a fixed order, followed by a preference question.

After: a randomised or rotated list, with the rotation method recorded.

The buyer does not need to become a questionnaire methodologist. They do need to ask whether order was tested and whether the supplier can show the final sequence.

What a wording change can do to a headline

Consider an illustrative UK consumer study on a subscription service. One version asks: ‘Would you agree that the monthly fee represents good value?’ A second asks: ‘How would you rate the monthly fee: very good value, fairly good value, neither good nor poor value, fairly poor value or very poor value?’ The first invites agreement with a proposition; the second asks for a judgement across a balanced scale. The second is not guaranteed to produce a lower score, but it gives the board a more interpretable distribution.

In a B2B setting, imagine a survey asking: ‘Are you involved in purchasing software?’ A positive answer may include anyone who uses the software and has an opinion. A role-grid alternative may show that only a minority can approve expenditure. The market has not changed between the two versions; the definition of involvement has.

These examples are not arguments that every number is fragile. They are arguments for knowing which sentence produced the number before the board treats it as a fact.

A short sign-off checklist for UK research buyers

Before approving a board-bound questionnaire, the buyer should be able to answer yes to the following:

• Is each question tied to a specific decision or reporting need?
• Does any question contain an adjective or claim that invites a particular answer?
• Has each question been checked for double-barrelled wording?
• Are timeframe, category and respondent role defined in plain English?
• Is the answer scale balanced and appropriate to the question?
• Has question order been tested or rotated where priming is plausible?
• Has the questionnaire been piloted with a small number of representative respondents?
• Is the final version, including any changes after pilot, retained in the project record?

A ‘no’ is not necessarily fatal. It is a prompt for a documented reason. The purpose is not to slow every commission; it is to make the judgement visible.

What to ask the supplier before the pack is locked

Request the final wording, not just the fieldwork dates and sample size. Ask what changed after the pilot and why. Ask how lists were rotated and how scales were presented. Ask whether any sensitive or leading phrasing was tested in alternative form. If the board is likely to act on a narrow margin, ask for a sensitivity check rather than a post-hoc defence.

At Jordan Research, the editorial view is straightforward: the questionnaire is a board-level artefact even when it is not a board-level document. A short wording audit before sign-off is a modest procurement step with a disproportionate effect on the credibility of the final number. The board may never read the sentence, but the sentence will still have shaped the statistic.