Concept testing methods, compared

Concept testing methods compared: monadic vs. sequential monadic vs. comparative, qualitative vs. quantitative, and how to run each.

Concept testing exposes people to an early idea, feature, positioning statement, price, or product before further development. The research design determines how many concepts each participant sees, in what order, and whether the study collects scores, open responses, or both. Poor choices on those points can make the results difficult to interpret. The main exposure formats are monadic, sequential monadic, and comparative; each can use qualitative and quantitative questions.

Monadic, sequential monadic, and comparative concept tests

These formats differ in how many concepts each participant evaluates and the context in which they see them.

Monadic testing shows each participant one concept. Participants do not see another concept to compare against, so their scores and reactions reflect the concept in isolation, similar to encountering a new product or feature without a lineup of alternatives. To compare concept A with concept B, run separate matched cells with the same screener, questions, and approximate sample size, then compare the aggregated results. Each cell needs its own participant group, so testing four concepts monadically requires four full samples.

Sequential monadic testing keeps the one-at-a-time judgment within one participant session. Each participant evaluates concept A, answers the reaction questions, then repeats the process for concept B and the rest of the set. One sample covers every concept, but the format introduces order effects. A concept's position can influence its rating, with the direction and size of that effect depending on the category. Rotate concept order across participants to reduce systematic bias toward one option.

Comparative testing, also called side-by-side testing, shows participants two or more concepts together and asks them to choose, rank, or allocate preference across the set. A forced choice or ranking gives a direct differentiation signal and needs a smaller sample than separate monadic cells. It does not measure how a concept performs in isolation. A concept that loses a side-by-side comparison can still score well alone, and the lineup may influence reactions through comparison and anchoring.

Qualitative concept tests vs. quantitative concept tests

Choose the exposure format separately from the response format. Scores and open responses answer different questions, and many concept tests use both.

A quantitative concept test asks every participant to rate fixed dimensions such as purchase intent, uniqueness, appeal, believability, or value. The resulting scores are comparable across concepts, segments, and later test rounds. Use this format when the decision depends on a measurable difference between concepts.

A qualitative concept test asks participants to explain the concept in their own words, identify confusing parts, and describe what would change their interest. Synthesis takes longer than reading a chart, but the responses can explain a rating and identify issues such as a confusing name, an unconvincing claim, or an irrelevant benefit. See User interviews vs. surveys: which one, when for the same qualitative and quantitative choice across research methods.

A common design places an open question immediately after a structured rating and asks what drove the score. One AI-moderated study can collect both responses in the same flow.

How many participants a concept test needs

Concept-test sample size follows the general principles in How many participants do you need for user research?. The exposure format also changes the total sample required.

Monadic testing needs the most total participants, since every concept requires its own adequately sized cell. Comparative and sequential monadic designs are more sample-efficient, because a single participant's session covers every concept, which is a large part of why those formats show up more often once you're screening more than two or three concepts and don't have the budget for a monadic cell each.

If you're comparing results across segments, existing customers versus prospects, or one persona versus another, size each segment as its own cell within whichever format you choose rather than pooling them. A concept that wins overall can still be losing badly with the one segment you actually care about, and pooled results will not show you that.

Before recruiting a full sample, use a small qualitative round or an AI-persona pass to reduce a long list to the two or three concepts that need a properly sized quantitative test.

Running concept tests in Versive

Versive supports an early pass with AI personas and a subsequent study with real participants for decision-making evidence.

For an early review, AI tests present an uploaded image, Figma prototype, or live page to one or more AI personas. They return reports, transcripts, screenshots, and prioritized findings in minutes without recruitment. Use this step to find unclear concepts before assigning them space in a real-participant sample. What are synthetic users, and when should you trust them? explains the limits: purchase intent and willingness to pay are real-world attitudes that require validation with actual participants.

For the real-participant test, mix structured and AI-moderated questions in one study. A Rating Scale, Star Rating, or NPS-style question records purchase intent, appeal, or uniqueness as a comparable number. An AI Question can then ask participants to explain the score, with follow-ups based on each response. For comparative or sequential monadic designs, add a Ranking or Allocation question after participants have seen every concept. See Survey question types, and when to use each for the structured and AI question types a study can mix.

Recruiting works the same way regardless of format: bring in your own users with a share link or an embedded flow, invite people already saved to your organization's audience, or draw on Versive's participant panel when you need a screened, demographic-matched sample you don't already have. If you're running separate monadic cells, use the same screener and target sample size on each to keep the cells comparable, and set a quota on each so no cell over-fills. Once responses are in, AI analysis turns the open answers into themes and quotes and charts the structured scores automatically, so comparing across concepts, or across monadic cells, doesn't mean exporting to a separate tool first.

Use monadic testing to measure a concept in isolation. Use comparative or sequential monadic testing to compare several concepts with a smaller total sample. Pair structured scores with open reactions when the decision requires both a ranking and the reasons behind it.