Survey & research glossary
Plain-language definitions for the terms you'll run into designing and analyzing surveys.
A/B test
Sending two versions of a survey, question, or invitation to different respondents to compare which one performs better.
Abandonment rate
The share of respondents who started a survey but left before submitting it - the inverse of completion rate.
Access code
A unique code given to a specific respondent so only they can open and complete a survey, commonly used for controlled panels or invite-only research.
Anonymous response
A survey submission collected without recording any information that could identify who submitted it.
API
Application Programming Interface - a way for other software to read your surveys and responses programmatically instead of through the web panel.
Audit log
A per-organization record of who did what and when - logins, permission changes, and other sensitive actions - kept for accountability and review.
Benchmark
A reference score, such as an industry-typical NPS or CSAT, used to judge whether your own result is strong, average, or weak.
Bias
A systematic distortion in survey results caused by how questions are worded, who was sampled, or how the survey was distributed - as opposed to random error.
Branching
Routing respondents to different questions or pages based on how they answered an earlier question. See also Skip logic.
CAPTCHA
A challenge used to block automated bots from submitting fake survey responses, distinguishing human respondents from scripts.
Census
Surveying every member of a population rather than a sample of it - practical only for small or fully known populations.
Closed-ended question
A question with a fixed set of answer choices, such as single choice, multiple choice, or a rating scale - as opposed to an open-ended text question.
Coding
Sorting open-text responses into a defined set of categories or themes so they can be counted and compared like closed-ended answers.
Completion rate
The share of respondents who started a survey and went on to finish it, usually expressed as completed responses divided by delivered invitations.
Learn moreConfidence interval
The range a survey result is likely to fall within across the true population, calculated as the result plus or minus the margin of error.
Confidence level
How sure you want to be that your confidence interval actually contains the true value - typically expressed as 90%, 95%, or 99%.
Consent
A dedicated question type used to record a respondent's explicit agreement before continuing - commonly required for research involving personal or sensitive data.
Constant-sum question
A question that asks respondents to divide a fixed number of points (such as 100) across several items to show relative priority.
Construct
An underlying concept a survey is trying to measure - such as "trust" or "satisfaction" - that can't be observed directly and is instead measured through a set of related questions.
Correlation
A statistical relationship between two variables, showing whether they tend to move together - not the same as one causing the other.
Cross-tabulation
Comparing answers to one question broken down by answers to another - for example, satisfaction scores split by customer segment - to reveal patterns a single-question view would hide.
CSAT
Customer Satisfaction Score - a metric measuring satisfaction with a specific interaction, usually on a 5-point scale from very dissatisfied to very satisfied.
Learn moreCustomer Effort Score
A metric measuring how much effort a customer had to put in to get something done, usually on a 1-7 scale from very low effort to very high effort.
Data cleaning
Removing or flagging low-quality responses - duplicates, speedsters, straight-liners - before analyzing results.
Data minimization
Collecting only the personal data actually needed for a survey's purpose, rather than asking for extra information "just in case."
Data retention
How long collected responses and personal data are kept before being deleted, typically governed by a documented policy.
Detractor
In an NPS question, a respondent who scores 0-6, indicating they are unlikely to recommend and may actively discourage others.
Learn moreDisplay logic
A rule that shows or hides a specific question based on an earlier answer, without necessarily skipping ahead or ending the survey.
Distribution channel
The method used to reach respondents - a public link, QR code, email invitation, or embedded form - each with different reach and response-rate characteristics.
Drop-off
Respondents who start a survey but abandon it before reaching the end, often concentrated at a specific question that's too long, sensitive, or confusing.
Duplicate response
A second or later submission from the same respondent to the same survey, which analysis should generally exclude or flag.
Encryption at rest
Protecting stored data - such as saved responses - so it stays unreadable if the underlying storage is ever accessed without authorization.
Encryption in transit
Protecting data while it travels between a respondent's device and the server, typically via HTTPS/TLS, so it can't be read if intercepted.
Finite population correction
A statistical adjustment that reduces the required sample size when your population is small enough that sampling a meaningful fraction of it improves precision.
Forced-choice question
A question with no neutral or "no opinion" option, requiring respondents to pick a side even if their true view is mixed.
Frequency distribution
A summary showing how many responses fall into each possible answer or category for a question.
GDPR
The General Data Protection Regulation - EU law governing how personal data, including survey responses that identify someone, must be collected, stored, and handled.
Identified response
A survey submission linked to a specific respondent's identity, as opposed to an anonymous response.
Likert item
A single statement rated on a Likert scale, such as "I am satisfied with the onboarding process" rated from strongly disagree to strongly agree.
Likert scale
A rating scale (commonly 5 or 7 points) that lets respondents indicate their level of agreement, satisfaction, or frequency, from one extreme to the other.
Longitudinal study
Research that surveys the same population repeatedly over time to track how attitudes or behavior change.
Margin of error
How much a survey result might differ from the true value across your whole population, expressed as a plus-or-minus percentage.
Learn moreMatrix question
A grid-style question that asks respondents to rate several items (rows) against the same set of answer options (columns) in one compact table.
Mean
The arithmetic average of a set of numeric answers - add them up and divide by the count.
Median
The middle value in a set of numeric answers when sorted from lowest to highest - less skewed by extreme outliers than the mean.
Metadata
Information collected alongside a response but not part of the questions themselves, such as submission time, language, or source.
Mixed methods
Research that combines quantitative data (closed-ended questions, scores) with qualitative data (open-text answers, interviews) in the same study.
Multi-factor authentication
Requiring a second verification step beyond a password - typically a time-based code - before a creator or administrator can sign in.
Nonresponse bias
Distortion that occurs when the people who don't respond to a survey differ systematically from those who do, skewing the results.
NPS
Net Promoter Score - a loyalty metric from 0-10 "how likely are you to recommend us" question, calculated as the percentage of Promoters minus the percentage of Detractors.
Learn moreOpen-ended question
A question that lets respondents answer in their own words rather than picking from fixed options, such as a short-text or long-text field.
OpenID Connect
An identity protocol built on OAuth 2.0 used for single sign-on, letting a user authenticate through their organization's existing identity provider.
Outlier
A response or value that differs sharply from the rest of the data, which can distort averages if not identified and handled deliberately.
Panel
A pre-recruited group of people who agree to take surveys on an ongoing basis, often used for longitudinal or repeated research.
Passive
In an NPS question, a respondent who scores 7-8 - satisfied but not enthusiastic enough to actively promote you.
Learn morePiping
Automatically inserting a respondent's earlier answer into the text of a later question, so a survey can reference their own previous input.
Population
The entire group you want to draw conclusions about, whether or not you actually survey every member of it - for example, "all customers" or "all employees."
Probability sampling
A sampling method where every member of the population has a known, non-zero chance of being selected, making results statistically generalizable.
Promoter
In an NPS question, a respondent who scores 9-10, indicating strong loyalty and a high likelihood of recommending you to others.
Learn moreQuota
A cap on how many responses are collected from a particular segment of respondents, used to keep a sample balanced across groups.
Randomization
Presenting questions or answer options in a different order to each respondent, to reduce bias caused by position (such as always picking the first option).
Ranking question
A question that asks respondents to order a list of items by preference, rather than rating each one independently.
Rating scale
A question type where respondents pick a point on a numeric or star scale to express intensity of opinion, such as 1-5 or 1-10.
Reliability
Whether a survey question or scale produces consistent results when repeated under the same conditions.
Representative sample
A sample whose composition mirrors the population it's drawn from closely enough that its results can be generalized with confidence.
Response bias
A tendency for respondents to answer inaccurately - due to social pressure, question wording, or a desire to appear a certain way - rather than giving their true view.
Response credit
A prepaid unit that a pay-per-response survey consumes once per completed response, instead of drawing against a subscription's monthly response limit.
Response rate
How many of the people you invited actually took part in your survey, most commonly measured as completion rate.
Learn moreSAML
Security Assertion Markup Language - an XML-based standard used for single sign-on, letting an identity provider authenticate a user to the platform.
Sample
The subset of a population that actually responds to your survey, used to estimate what the whole population thinks.
Sample size
The number of respondents needed for a survey's results to reliably represent the target population, given a chosen confidence level and margin of error.
Learn moreSampling error
The natural difference between a sample's result and the true value for the whole population, which shrinks as sample size grows.
SCIM
System for Cross-domain Identity Management - a standard that lets an identity provider automatically create, update, and deactivate user accounts.
Sentiment analysis
Classifying open-text responses as positive, neutral, or negative in tone, typically done automatically rather than by manual review.
Skip logic
A rule that ends a survey early, or skips later questions, based on how a respondent answered an earlier one.
Slider question
A question answered by dragging a handle along a numeric range, rather than picking from discrete options.
Standard deviation
A measure of how spread out numeric answers are around the mean - a small standard deviation means most answers were close together.
Start rate
The share of people who received a survey invitation and actually opened and began the survey, regardless of whether they finished it.
Learn moreStatistical significance
A measure of how likely it is that a result reflects a real effect rather than random chance, typically based on sample size and the size of the observed difference.
Stratified sampling
Dividing a population into subgroups (such as region or department) and sampling from each one, to make sure smaller groups aren't underrepresented.
Survey fatigue
Declining response quality or willingness to participate caused by surveys that are too long, too frequent, or poorly targeted.
Survey logic
Any rule that changes what a respondent sees next based on their previous answers - includes skip logic, branching, and piping.
Survey weighting
Adjusting how much each response counts in the final results so the sample better reflects the true population, correcting for over- or under-represented groups.
Thank-you page
The page or message shown to a respondent immediately after they submit a survey, which can be customized with a message or a redirect to another URL.
Validity
Whether a survey question actually measures the concept it's intended to measure, as opposed to something else entirely.
Webhook
An automated HTTP request sent to a URL you register, fired the moment an event happens - such as a respondent completing a survey - so external systems can react immediately.
Weighted average
An average that accounts for how many responses fall at each point on a scale, rather than just counting how many chose the top options - used alongside CSAT to show the full distribution.
Learn moreZ-score
A statistical value corresponding to a chosen confidence level (for example 1.96 for 95% confidence), used in the formulas for margin of error and sample size.
Clear definitions improve survey decisions
Terms such as anonymous, representative, significant, confidential, compliant, and secure carry specific expectations. Use them only when your survey method and implementation actually support the claim.
Know the terms, now try the platform
Create a free workspace and put these definitions to work in a real survey.