Survey research: from questionnaire design to weighted results you can act on

Questions this report answers

• What share of the target group uses, knows or would buy the product?
• How do results differ by age, region, income or customer segment?
• Which features or messages matter most, and to whom?
• Has satisfaction, awareness or intent really changed since the last wave?
• How many respondents are needed for reliable subgroup results?
• Which survey mode reaches this audience with acceptable data quality?

When you need it

Surveys are the right tool when a decision depends on a number: how large a segment is, how many customers would switch, whether a shift in awareness is real or noise. Typical uses include sizing demand from stated behaviour, customer satisfaction and NPS tracking, brand awareness waves, pricing research with Van Westendorp or Gabor-Granger questions, and post-launch measurement.

A survey needs something clear to measure. If you do not yet know which options, words or concerns matter to people, start with qualitative work such as focus groups or interviews, then use what you learn to write answer lists that reflect how respondents actually think. Working in that order also keeps the questionnaire shorter.

Surveys are weak at explaining why people feel as they do, at predicting reactions to products respondents cannot picture, and at reaching very small, hard-to-find groups. When the population is a few dozen key accounts, interviewing all of them usually beats sampling. And when the behaviour itself can be observed — in sales data, web analytics or a field experiment — measuring it directly is more reliable than asking about it.

Sample table of contents

1. Executive summary
   key findings, each with its margin of error
2. Objectives and hypotheses
   what the survey was designed to measure
3. Methodology
   mode, sample source, size, quotas, fieldwork dates, weighting
4. Sample profile
   achieved sample against population targets
5. Core results
   question-by-question findings with charts
6. Segment comparisons
   statistically significant differences between groups
7. Driver analysis
   which factors relate most strongly to the main outcome
8. Trend view
   comparison with earlier waves or benchmarks, where available
9. Conclusions
   implications for the decision
10. Appendix
    questionnaire, data tables and technical notes

Methodology and sources

Questionnaire design

Every question maps to an objective. We screen wording for leading phrasing and double-barrelled items, keep scales consistent and pilot the script before full launch. Long online questionnaires lose attention and data quality, so we check timing with the survey length calculator and cut what no decision depends on.

Choosing the mode

CAWI (online) is fastest and suits most populations with high internet use; CATI (telephone) reaches less connected groups and business respondents; CAPI (face to face) works for long interviews, product handling and areas with weak connectivity. Coverage varies by country. Eurostat puts daily internet use at 98.4% of individuals in the Netherlands in 2025, against 82.7% in Bulgaria and 89.5% across the EU, so an online-only survey of Bulgarian adults leaves more people out and often needs a telephone or face-to-face top-up. For a handful of questions, an omnibus survey shares fieldwork with other clients and reports faster.

Sample and weighting

Sample size follows the precision you need: for a large population at 95% confidence, about 385 completes give a margin of error of roughly ±5 percentage points. Plan by subgroup, not just in total, with our sample size calculator. Quotas on age, gender and region come from official population data, and results are weighted to correct the imbalances that remain.

Quality checks

Speeders, straight-liners and failed attention checks are removed, open answers are coded, and a second analyst checks every table against the raw data.

What you receive

• Final questionnaire and a test link for your approval
• Topline results soon after fieldwork closes
• Full report with charts, significance testing and recommendations
• Cross-tabulations in Excel for all agreed segments
• Clean, weighted respondent-level data (SPSS, CSV or Excel) with codebook
• Presentation and Q&A session with your team

How the project runs

1. Briefing and questionnaire drafting — 3–5 days
2. Your review, translation and scripting — 3–5 days
3. Soft launch and fixes — 1–2 days
4. Fieldwork — 1–3 weeks, depending on mode and incidence
5. Data cleaning, coding and weighting — 2–4 days
6. Analysis, report and presentation — 3–7 days

Common mistakes and how to avoid them

• Writing questions before objectives. If nobody can say what they would do with an answer, cut the question.
• Sizing the sample for the total only. A sample that works overall can leave key subgroups too thin; set quotas per segment.
• Ignoring mode effects. People answer differently online and to an interviewer, especially on sensitive topics; keep the mode constant across tracking waves.
• Reading stated intent as sales. Purchase-intent scores overstate real buying; calibrate them against past launches or behavioural data.
• Skipping the soft launch. A small first batch catches broken routing, unclear wording and timing problems before the full sample is used up.

Who it suits

Updated: 5 Nov 2026