Glossary
Market research terms, in plain English
The vocabulary you actually run into when you field a study — defined without the jargon that usually comes with it.
- Net Promoter Score (NPS)
- A loyalty metric from a 0–10 “how likely are you to recommend” question. Respondents are grouped into promoters (9–10), passives (7–8), and detractors (0–6); the score is the promoter percentage minus the detractor percentage, ranging from −100 to +100.
- CSAT (Customer Satisfaction)
- A satisfaction measure, usually the percentage of respondents who rate an experience positively (e.g. 4–5 on a 5-point scale). Best used for a specific, recent interaction rather than overall sentiment.
- CES (Customer Effort Score)
- Measures how much effort a customer had to expend to get something done. Lower effort correlates strongly with loyalty, which makes CES a useful complement to satisfaction and NPS.
- MaxDiff (Maximum Difference Scaling)
- A method where respondents repeatedly pick the best and worst item from small sets. It forces trade-offs and produces a clean preference ranking, avoiding the “everything is important” flatness of rating scales.
- Conjoint analysis
- A choice-based method that shows respondents full product profiles and asks them to choose. By analyzing the choices, you estimate the value (utility) of each attribute and level — useful for pricing and feature prioritization.
- Crosstab (cross-tabulation)
- A table that breaks one question’s results down by the categories of another (for example, satisfaction by plan tier). Crosstabs are the workhorse of survey analysis for spotting differences between segments.
- Significance testing
- Statistical tests (such as t-tests or chi-square) that estimate whether a difference between groups is likely real or just sampling noise. A result is “significant” when it is unlikely to have occurred by chance alone.
- Weighting
- Adjusting a sample so it matches known population proportions (age, gender, region). Weighting corrects for over- or under-representation so results generalize more accurately to the target population.
- Sample
- The subset of a population you actually survey. A representative sample mirrors the population on the characteristics that matter, so findings can be generalized with a known margin of error.
- Panel
- A managed pool of pre-recruited respondents you can buy sample from, often filtered to specific demographics or behaviors. Providers like Cint and Prolific supply panel for studies that need a targeted audience.
- Screening / screener
- Early questions that qualify or disqualify respondents so only your target audience continues. Screened-out respondents are routed out before the main survey, protecting data quality and budget.
- Quota
- A cap on how many respondents from a given group you accept (for example, 50% women, 50% men). Quotas keep the final sample balanced and can be independent or interlocked across multiple variables.
- Skip logic / branching
- Rules that change which questions a respondent sees based on earlier answers. Good logic keeps surveys short and relevant, which improves completion rates and data quality.
- Open-ended question
- A free-text question with no preset answers. Rich but often shallow unless probed; adaptive follow-ups can turn a vague answer into a specific, analyzable one.
- Margin of error
- The range within which the true population value is likely to fall, given your sample size and confidence level. Larger samples produce a smaller margin of error and more precise estimates.