Platform
Features
Solutions
Research Glossary
Research can sometimes feel like its own language reserved for those in the know. We've put together a growing list of terms in the space to keep you in the know.
Research Glossary
Research can sometimes feel like its own language reserved for those in the know. We've put together a growing list of terms in the space to keep you in the know.
Research Glossary
Research can sometimes feel like its own language reserved for those in the know. We've put together a growing list of terms in the space to keep you in the know.
A
A/B testing compares two versions through random assignment and planned outcome analysis to estimate the effect of a change, including uncertainty.
Accessibility concerns whether people with disabilities can perceive, understand, navigate, and use a product, including through the assistive technologies and settings they rely on.
An activation metric tracks an early user or account milestone intended to indicate that someone has begun receiving a product’s core value.
Ad testing evaluates advertising creative to understand how it is noticed, interpreted and connected with a brand or intended response.
Affinity mapping groups related observations or ideas to help researchers examine patterns, compare interpretations and organise material for further analysis.
An affordance is an action made possible by the relationship between a person and an object or environment; a signifier communicates where or how an action is available.
Agile research integrates user research with ongoing product development, using appropriately scoped studies to inform decisions as the team learns and delivers.
An AI research assistant supports tasks such as study preparation, retrieval, transcription, and analysis, with researchers checking its outputs against the relevant evidence.
An AI-moderated interview uses software to ask research questions and adapt follow-ups to a real participant’s responses, within a study designed by researchers.
Anonymity means a research contribution cannot reasonably be linked to an identifiable person in the relevant context; confidentiality instead concerns protecting information that may identify them.
Attrition rate is the proportion of a defined study group lost before a specified follow-up or completion point, with the pattern of loss affecting how results can be interpreted.
B
A baseline measurement records a starting condition before a change, providing a documented reference for interpreting later results.
Behavioral analytics examines recorded product actions to understand journeys, adoption, retention and differences between users or accounts.
Behavioural design uses evidence about how people act to shape products and services, examining the context, obstacles, prompts, and consequences surrounding a specific behaviour.
Benchmarking compares performance with a defined reference, such as an earlier product version, a competitor or a relevant external dataset.
Beta testing puts a developing product into the hands of external users to learn how it performs in realistic use before a broader release.
Brainstorming is a structured activity for generating possible ideas around a question, separating initial exploration from later evaluation and testing.
Brand awareness concerns whether people recognise or recall a brand within a relevant market, category or buying situation.
Brand tracking measures a brand’s awareness, associations and other relevant indicators over time using a consistent research approach.
Business intelligence brings organisational data together for reporting and analysis, helping teams understand performance, investigate patterns, and make decisions using defined measures.
C
Card sorting asks people to group information into categories, helping researchers explore relationships and labels when designing an information architecture.
Causal research investigates whether and how an intervention or factor changes an outcome, using a design and assumptions that address alternative explanations.
Churn analysis investigates which customers stop using or paying for a product, when they leave and what evidence may explain the departure.
Clickstream analysis examines ordered records of digital interactions to understand the routes people take through a website or product.
A cognitive bias is a systematic tendency in attention, memory or judgement that can shape how people interpret information and make decisions.
Cognitive load is the demand a task places on mental processing, including remembering information, interpreting choices, and coordinating the steps needed to reach a goal.
Cohort analysis compares outcomes for groups with a shared characteristic or starting event, often tracking their behaviour at equivalent times after signup.
Competitive analysis examines the alternatives people use to meet a need, comparing their offers and experiences to inform product or market decisions.
Comprehension testing examines how people interpret and apply information. It uses explanations and realistic scenarios to reveal misunderstandings that clarity ratings or preferences can miss.
Concept testing explores how an intended audience understands and responds to a proposed product, feature or experience before committing to its full development.
A confidence interval is a range calculated from data by a method with a stated long-run coverage rate, expressing uncertainty around an estimate under its assumptions.
Conjoint analysis estimates preferences for product attributes by asking people to evaluate alternatives that combine those attributes in different ways.
A consumer panel is a maintained group of individuals or households contributing information for research, such as purchase records, survey responses or product-use feedback.
Content analysis systematically examines communication by organising material into categories and interpreting its characteristics or meaning, quantitatively or qualitatively.
Contextual inquiry combines observation with questions during an activity, helping researchers understand how people work within their actual circumstances.
Continuous discovery is an ongoing practice of customer contact and small research activities that helps a product team pursue a desired outcome and decide what to build.
Correlation describes a statistical relationship; causation concerns the effect of changing a factor compared with an appropriate alternative. Association alone does not establish that effect.
Customer discovery investigates prospective customers’ problems, circumstances and buying decisions to test the assumptions behind a product or business.
Customer Effort Score (CES) measures reported effort or ease in accomplishing a particular goal; interpretation depends on the question, scale direction and scoring convention.
Customer experience is how a person experiences a company across interactions and over time, including expectations, use, support, payment, and the consequences of those encounters.
Customer journey mapping brings evidence about a person’s actions, questions, and experience across a defined process into a shared view, exposing gaps and opportunities for improvement.
Customer Satisfaction Score (CSAT) summarises reported satisfaction with a defined experience, product or relationship, using an explicitly stated scale and calculation.
d
Data cleaning identifies and handles errors, inconsistencies and quality issues through documented checks and transformations that support credible analysis.
Decision fatigue describes a proposed change in decision-making after sustained demands, such as greater avoidance or reliance on defaults; its causes and generality require careful interpretation.
Demographic research examines population characteristics such as age, location, household composition, and education, helping describe groups and investigate relevant differences in experience.
A design sprint is a time-limited process for exploring a focused challenge, choosing a solution, building a prototype, and learning from customer research.
Design thinking is an approach to solving problems through understanding people, exploring alternatives, making ideas tangible, and testing what those ideas enable.
A diary study asks participants to record relevant experiences over time, often close to when they occur, to investigate routines, context and change.
Dogfooding means using your organisation’s own product internally to discover problems and improve it through everyday work.
A double-blind study commonly withholds condition information from participants and relevant researchers, but a clear report specifies exactly which roles were blinded and how.
Driver analysis examines which measured factors are associated with an outcome, using methods such as correlation, regression or relative-importance analysis.
e
Empathy mapping organises evidence about a person’s perspective, often around what they say, think, do and feel, while making gaps and assumptions visible.
End-user research studies the people who directly use a product or service to accomplish tasks, distinguishing their experience from that of buyers or other representatives.
Ethnographic research develops a contextual understanding of a group, practice or social setting through fieldwork, observation, participation and other relevant evidence.
Experimental design specifies the conditions, assignment, outcomes and procedures used to investigate the effect of an intervention.
Exploratory research investigates a problem space that is not yet well understood, developing understanding, candidate explanations and more precise research questions.
Eye tracking estimates where a person’s eyes are directed over time, helping researchers investigate visual search, scanning and gaze within an interface or environment.
f
Feature prioritisation is deciding which product changes deserve attention first, using evidence about user needs, expected outcomes, effort and constraints.
A feedback loop uses information about the result of an action to guide subsequent action, with a defined way to observe, interpret, respond, and check what changed.
First-click testing records where people first click or tap to begin a task, helping researchers evaluate whether an interface suggests useful starting points.
A five-second test briefly shows a design before asking participants what they recall and understand, helping assess its immediate message and impression.
A focus group is a facilitated discussion in which participants explore a topic together, allowing researchers to examine both their accounts and the interaction between them.
Formative research informs the development or improvement of a product or service, providing evidence that helps shape the next design decision.
g
Gamification in research adds game-like elements to participation or study materials, requiring evaluation of how those elements affect both engagement and the evidence collected.
Generalizability concerns how far research findings apply beyond the participants, conditions and period actually studied.
Grounded theory is a qualitative methodology that develops an explanatory account through iterative data collection, comparison, coding, and theoretically directed sampling.
Guerrilla research is a lightweight, opportunistic approach to gathering feedback or observing use, usually through short sessions with people recruited in an accessible setting.
H
The Hawthorne effect commonly refers to behaviour changes associated with being observed or studied; the mechanisms and size of those changes depend on the research setting.
Heatmaps use colour to show the distribution of recorded activity, such as clicks, scroll reach or gaze. The type of data determines what the map can reveal.
A heuristic evaluation is an expert inspection of an interface against established usability principles, producing reasoned assessments of likely problems for further investigation or correction.
Hypothesis testing evaluates data against a specified statistical hypothesis using a chosen model and decision procedure.
I
An in-home usage test asks people to use a product in their everyday environment over an agreed period and provide evidence about the experience.
An in-product survey asks questions within a product experience, using its placement and timing to reach people in a relevant context while accounting for who is excluded.
Information architecture organises, structures and labels content so people can understand relationships and find the information they need.
Informed consent is a person’s voluntary agreement to participate in research after understanding what the study involves, how their contribution will be used, and their choices.
Insight mining examines existing research and feedback for evidence relevant to a current question, preserving the original context and identifying where new research is still needed.
An intercept survey recruits people during an activity or visit and asks questions in that context, either in a physical setting or through a website or product invitation.
An interview guide sets out a research session’s topics, questions, and practical arrangements, helping the interviewer explore experiences without losing focus.
Iterative research proceeds through successive rounds, using what is learned to refine a design, research question, or study before the next round.
J
Jobs to be Done examines the progress people seek in particular circumstances and how they choose products, services, or other alternatives to help them achieve it.
K
The Kano model classifies how product attributes relate to customer satisfaction, distinguishing basic expectations, performance benefits and attractive extras.
A key performance indicator (KPI) is a measure selected to assess progress towards a specific objective and guide decisions about performance.
Knowledge management in research is the practice of preserving, organising, maintaining, and sharing research evidence so it can inform decisions beyond the original study.
L
Lean UX is a collaborative approach to product design that makes assumptions explicit and uses research and experiments to pursue meaningful changes in user behaviour.
A Likert scale combines responses to related agreement items to measure an attitude or construct. A single agreement question is more precisely called a Likert item.
A longitudinal study repeatedly observes the same individuals, accounts or other units over time to investigate change and the sequence of experiences.
M
Machine learning in research uses models that learn patterns from data to support tasks such as classification, transcription, clustering, and prediction, with evaluation suited to each use.
The margin of sampling error commonly expresses the half-width of a symmetric confidence interval; it describes sampling uncertainty rather than every possible survey error.
Market research investigates customers, demand, alternatives, and the conditions surrounding a market to inform decisions about an offering and how it reaches people.
MaxDiff is a survey method that asks people to choose the best and worst, or most and least important, items from repeated sets to estimate relative preferences.
A mental model is a person’s understanding of how something works, used to predict outcomes and decide how to act.
Message testing investigates how an intended audience understands and responds to communication, including its relevance, credibility and implications.
Microcopy is the short functional text in an interface, such as labels, hints, errors, and confirmations, that helps people understand an action and its consequences.
Mixed-methods research intentionally integrates qualitative and quantitative approaches so their combined evidence addresses a shared research purpose.
Moderated testing is a research session in which a researcher observes a participant using a product or prototype and asks questions to understand the experience.
Moderator bias is systematic influence from a researcher’s wording, reactions, assistance or interpretation that affects a participant’s behaviour or the resulting findings.
Monadic testing asks each participant to evaluate one concept, message or product. When studying alternatives, researchers compare responses from separate groups rather than showing competing versions to the same person.
N
Net Promoter Score (NPS) subtracts the percentage of detractors from promoters on a 0–10 likelihood-to-recommend question, producing a score from −100 to +100.
Non-response bias occurs when people who answer a study differ from those who do not in ways that affect the result being estimated.
A north star metric is a principal product measure intended to reflect customer value and align team decisions, supported by input measures and checks on trade-offs.
O
Observational research systematically examines behaviour, events or conditions. In an observational study design, researchers do not assign the intervention or exposure of interest.
Open-ended questions let participants answer in their own words, revealing experiences, explanations and language that prepared response options may miss.
An opportunity solution tree connects a desired product outcome with researched customer needs, possible solutions, and tests of the assumptions behind those solutions.
Outliers are observations unusually distant from other values or an expected pattern. They may reflect errors, valid extremes or a mismatch between the data and the model.
Overgeneralization extends a research finding beyond the people, situations, time or level of certainty that its evidence supports.
P
Packaging testing examines how people find, interpret, choose and use a pack, using research suited to its visual and physical functions.
Panel research uses an established participant group, either as a recruitment source for studies or to follow the same units repeatedly over time.
Participant observation combines taking part in an activity or social setting with systematic observation and interpretation of its practices and relationships.
Participant recruitment is the process of finding and involving people whose experience is relevant to a research question. It includes eligibility criteria, recruitment channels, screening and participation arrangements.
The path to purchase describes the events, influences and decisions through which someone moves towards buying a product or service.
Persona development turns research into a small set of evidence-based representations of relevant user patterns, helping teams reason about goals, behaviour, and context.
A pilot study is a smaller run used to assess and refine the feasibility, procedures or instruments of a planned research study.
Predictive analysis uses data and statistical or machine-learning models to estimate unknown outcomes, often future events such as demand, conversion or cancellation.
Preference testing asks participants to choose between design alternatives and explain their choice, providing evidence about their stated reactions.
Primary research collects original evidence for a research purpose, through activities such as interviews, surveys, observation, or experiments designed for the question at hand.
Probability sampling selects units through a random mechanism with known, non-zero inclusion probabilities, supporting population estimation under the study’s design and assumptions.
Product-market fit is the extent to which a product meets a meaningful need in a market, evidenced by customers choosing it and continuing to find it valuable.
Prototype fidelity describes how closely a prototype represents the intended product in appearance, content, interaction and the extent of the experience it includes.
Prototype testing asks people to use a representation of a product to investigate design questions before the complete product is built.
Purchase intent is a person’s stated likelihood of buying a specified product or service, usually measured under described conditions in a research study.
Purposive sampling deliberately selects participants or cases for experience and characteristics that can illuminate the research question.
Q
Qualitative research examines experiences, meanings, practices and context through material such as interviews, observations, documents and images.
Quantitative research uses numerical measurement and analysis to describe patterns, estimate quantities, compare conditions, or examine relationships under a defined research design.
Questionnaire design is the process of developing questions, answer options, sequence and presentation so a study collects information people can understand and answer accurately.
R
Random sampling uses a specified chance mechanism to select units from a population or sampling frame, providing a basis for inference under the sampling design.
Rating scales provide ordered response options for expressing an evaluation, attitude, intensity or frequency. Their meaning depends on the question and the response labels.
Reliability concerns the consistency of measurement under specified conditions, such as repeated occasions, different raters or related scale items.
Remote testing evaluates a product or prototype with participants in another location, through live moderated sessions or independently completed research activities.
A representative sample supports an account of a defined population on characteristics or outcomes relevant to the research question; recruitment and analysis determine that claim.
Research bias is systematic error arising from a study’s design, conduct, analysis or reporting that can distort its findings.
Research ethics concerns the responsibilities researchers have to participants and to the people who rely on their findings, throughout a study and its subsequent use.
ResearchOps provides the people, processes, and infrastructure that support research, including participant management, governance, tools, and the organisation of research knowledge.
Response bias is systematic distortion in answers arising from factors such as question wording, recall, response options and the circumstances of the study.
S
Sampling bias arises when coverage or participant selection systematically distorts a sample in ways relevant to the research question.
Screeners are recruitment questions used to assess whether a prospective participant has the characteristics or experience needed for a particular research study.
Secondary research examines data and findings that already exist, assessing their relevance and limitations for the question now being investigated.
Segmentation divides a market or user population into groups whose differences are relevant to a research, product or business decision.
Semi-structured interviews use prepared topics with flexible follow-up questions, allowing researchers to compare experiences while exploring unexpected details.
Sentiment analysis classifies expressed evaluation in text, often as positive, negative or neutral, sometimes distinguishing the product aspects being evaluated.
Session replay reconstructs a recorded visit from captured interactions and interface states, helping teams investigate behaviour around errors or difficult journeys.
Shadowing follows a participant through a period of activity, observing its sequence, setting and interruptions to understand how the work unfolds.
The Single Ease Question (SEQ) is a seven-point rating of how easy or difficult a task felt, collected immediately after an attempt. It measures perceived ease at task level.
Snowball sampling recruits participants through referrals from existing participants or contacts, helping researchers reach relevant people beyond conventional recruitment lists.
A result is statistically significant when its p-value meets a specified threshold under a statistical model; this does not establish the effect’s size or practical importance.
A survey collects information from a defined group through questions, using a consistent approach to examine reported experiences, behaviour and attitudes.
Survey fatigue is declining attention or willingness to participate when a questionnaire, or repeated survey invitations, places too much demand on respondents.
Synthetic users are AI-generated simulations of people or audiences that produce artificial responses or actions; their outputs require validation before supporting claims about real users.
The System Usability Scale (SUS) is a ten-item questionnaire that combines five-point agreement responses into a 0–100 score of perceived usability.
T
A target audience is the group a product, message or research study is intended to address, defined by characteristics relevant to its purpose.
Task analysis examines the actions, decisions, information and dependencies involved in achieving a goal, including work that happens outside a product.
A task scenario describes a realistic situation and goal for a research participant, leaving them to work out how to achieve it in the product being tested.
Task success rate, also called task completion rate, is the percentage of valid task attempts that meet a predefined success criterion. It measures completion under the conditions of a usability study.
Thematic analysis develops and interprets patterns of meaning across qualitative material, using an explicit analytical approach to address a research question.
A think-aloud protocol asks participants to describe their thoughts while attempting a task, adding a verbal account of expectations and decisions to observed behaviour.
Tree testing evaluates whether people can find information in a proposed navigation hierarchy, using realistic tasks and a text version of the structure.
Trend analysis examines measurements over time to understand sustained direction and changes, distinguishing them from seasonal patterns, short-term variation, and changes in measurement.
Triangulation examines a research question through multiple methods, sources, investigators or perspectives, comparing what each contributes and where findings differ.
U
UMUX is a four-item perceived-usability questionnaire; UMUX-Lite is a related two-item version. Each requires its own documented wording and scoring procedure.
Unmoderated testing lets participants attempt research tasks independently, with instructions and tools collecting their responses or behaviour without a researcher present.
Usability testing observes people attempting realistic tasks with a product or prototype to understand difficulties and evaluate how well the design supports their goals.
A user flow represents the steps, decisions and system responses involved in achieving a particular goal within a product.
User interviews are research conversations about people’s experiences, decisions, and circumstances. They help teams understand the context behind a need or behaviour.
User onboarding is the experience of helping new users understand a product, complete necessary setup and begin accomplishing something valuable with it.
User research systematically studies people’s goals, behaviour, experiences and circumstances to inform decisions about products and services.
User-centered design develops and evaluates products through an explicit understanding of users, their tasks and circumstances, with research informing iterative decisions.
UX metrics quantify aspects of an experience, including observed task outcomes and reported perceptions, to inform decisions about product quality.
V
Validity concerns whether evidence supports the interpretation and conclusions drawn from a measure or study, for its intended purpose and context.
Value proposition testing examines whether an intended audience understands, values and believes the benefit a product promises.
Voice of the Customer is a programme for collecting, interpreting and acting on customer feedback across research and everyday interactions.
W
Web analytics measures and analyses recorded website activity, including visits, events, traffic sources and conversions, to investigate patterns of use.
Willingness to pay is the maximum amount someone would exchange for a particular product or service under specified circumstances.
Win/loss analysis investigates sales decisions to understand why prospective customers chose an offer, selected an alternative or decided not to buy.
Wizard of Oz testing uses a human operator to simulate part or all of a system’s behaviour, allowing researchers to study an interaction before the capability is built.
Word association asks people to respond to a stimulus with the words or phrases that come to mind, helping researchers explore immediate impressions.
X
Y
Yes/no question bias is an informal term for distortions that can arise when question wording or binary answers fail to represent a participant’s experience accurately.
Z