Concept Testing

Concept testing is a research step that puts a described idea in front of the people you hope will buy it, before anyone builds it. It answers one question: does anyone actually want this? You get a reaction, not a price on each feature.

You are testing a description, not a product. A paragraph, a sketch, a landing page or a one-minute video is enough, because the point is finding out whether the idea deserves to be built at all.

That makes it cheap and early — concept testing sits well before prototypes and further still before launch, at the stage where changing your mind costs a week rather than a quarter.

Key takeaways

• Concept testing puts a described idea in front of target buyers to see whether it lands, before anyone builds it.
• It produces a pass, a fail or a fix list; it does not put a value on individual features.
• Concept testing asks whether the idea lands, while conjoint analysis asks what each part of it is worth.
• Small samples work when the signal is strong, because you are looking for a clear reaction rather than a decimal point.

How it works

You write the concept in plain language, show it to people who match your target buyer, and ask a fixed set of questions. Same wording every time, so answers from different groups stay comparable.

Concept testing is one loop inside design thinking, which cycles through understanding, ideas and testing instead of running straight at a build.

Human-centered design makes the same case from the other direction: you start from a real person’s problem, so the concept you test is already a response to something.

Government teams formalise the habit. The UK’s Service Manual describes an alpha phase spent building prototypes to test your riskiest assumptions, deliberately before committing to a full build.

Digital.gov, the United States guidance site for government digital services, argues for mixing methods, noting that surveys, analytics and interviews support informed product decisions.

A beta testing program is not the same thing. Beta puts a working build in front of real users — concept testing puts a description in front of people who have never seen anything.

Existing customer sentiment is a cheap first filter. If the complaint your concept solves never appears in support tickets or reviews, find out why before you spend on fieldwork.

Plenty of teams run the fieldwork through a partner. Product outsourcing covers the build side, while recruitment, moderation and transcription are usually the first tasks to go out.

Watch how you word the concept. Describe the benefit and the cost together, because a description with no price attached collects enthusiasm that evaporates the moment money appears.

Recruit the right people, too. Twenty responses from your target buyer beat two hundred from a general panel — the second group will happily rate an idea they would never purchase.

Examples

Concept tests scale from a dozen conversations to a few hundred survey responses. The format changes with the budget, but the job stays the same: show the idea, capture the reaction, and decide whether to keep going.

A food manufacturer describes three new flavours in a paragraph each and shows them to 300 shoppers. Two get polite shrugs and one gets “when can I buy it”, which settles the question.

UK government teams run concepts through alpha as throwaway prototypes, building the least they need to kill a weak idea well before any delivery budget has been committed to it.

A business process outsourcing provider pitching a new managed-service line tests the written description on twelve existing clients before hiring a single team leader.

Software teams use a landing page. You describe the product, count how many visitors click through to a waitlist, and read the drop-off as the reaction it is.

Retailers test private-label ideas the same way. A described product, a shelf-price range and a photograph are enough to sort three candidate lines into one worth developing and two worth dropping.

Service businesses test the pitch, not the product. A clinic describing a new subscription plan to fifty existing patients learns whether the idea reads as good value or as a price rise in disguise — before reception has to explain it.

Related terms

Concept testing sits inside a family of research and design terms. Some describe the wider method it belongs to, one describes the later stage it is often confused with, and two cover who runs the work.

• Beta Testing Program: the structured trial of a working build with real users before launch.
• Design Thinking: the problem-first method that cycles through understanding, ideas and testing.
• Human-Centered Design: the practice of designing outward from a real person’s needs.
• Post-Call Survey: the short questionnaire sent straight after a service interaction.
• Innovation Outsourcing: the use of external teams for research, prototyping and new-product work.

FAQ

When should you run concept testing?

As soon as the idea can be described in a paragraph and before any engineering time is committed. Running it after a build turns useful research into expensive reassurance.

How many people do you need?

Twenty to fifty for qualitative reads, a few hundred if you want to compare concepts numerically. Strong signals show up early, and weak ones rarely improve with sample size. Recruitment quality beats volume every time.

What is the difference between concept testing and conjoint analysis?

Concept testing asks whether the idea lands at all. Conjoint analysis asks what each part of it is worth by making respondents trade attributes against each other.

Can you test more than one concept at a time?

Yes, and comparing three is usually more informative than testing one, since people find it easier to rank options than to judge a single idea in isolation.

Does concept testing predict sales?

No, it predicts interest, and interest is a useful early filter rather than any kind of revenue forecast.

Buyers looking for a partner to run the fieldwork can start with the vetted research providers in the Outsource Accelerator directory.