What is concept testing?
Concept testing is research done on an idea before money is spent making it real. A campaign concept, a product idea, a new name, a piece of packaging or a positioning line is shown to a sample of the people it is meant for, usually as a short description, a mood film, a storyboard or a rough mock-up, and they are asked what they understand, what they feel, whether they would buy and what they would change. The answers tell the team which of several directions is strongest and what each one needs before it goes into production.
It is distinct from testing the finished work. Pre-testing a completed film checks execution; concept testing checks the idea itself, while it is still cheap to change. The craft is in showing people enough to react to without showing them so much that they are reviewing the production rather than the thought.
Why it matters
Why concept testing matters
Because the most expensive mistakes in marketing are made at the idea stage and discovered at the launch stage. A campaign that rests on a concept the audience misreads will be beautifully shot, expensively placed and quietly ignored. Testing the concept first costs a fraction of the production and catches the misreading while the fix is a conversation rather than a reshoot. It also arms the agency: a concept with research behind it survives the client's internal reviews far better than one defended on taste alone.
The caveat is that audiences are poor at predicting what they will love and good at explaining what they do not understand. Concept testing is strongest as a diagnostic, finding confusion, offense or indifference, and weakest as a prediction of a hit. Used that way it sharpens brave ideas. Used the other way it sands them down to the one nobody objected to.
In practice
How concept testing works in practice
- 01
The method depends on the stakes. At the light end, a qualitative round: a handful of focus groups or one-to-one interviews, each shown two or three concepts as boards or animatics, moderated by someone who can get past the first polite answer. At the heavier end, quantitative surveys with several hundred respondents per concept, scoring appeal, relevance, uniqueness, believability and purchase intent against category norms. Many teams do both, the qualitative to understand and the quantitative to decide.
- 02
The stimulus matters. Concepts should be presented at the same level of finish so the research compares ideas rather than production values, and in a form that reflects where the idea will live: a vertical mock for a social idea, a board sequence for a film. Results come back as a readout the strategist and creative director read together, looking for the concept that is both strong and fixable, and for the specific line, image or claim that caused the trouble.
- 03
Digital advertising has added a faster version: putting several rough executions into a live platform with a small budget and letting real behavior, not stated intent, pick the winner. It is cheaper and more honest than a survey, though it tests what people click on rather than what they understand, which is not always the same thing.
- 04
The order of the questions decides what the research is worth. Respondents are asked what they understood, in their own words, before they are asked whether they liked it, because liking is easy to say and comprehension is what the brief needs. A playback question, asking them to tell a friend about it, shows whether the idea travels. The decision rules are agreed before the results arrive: what finding kills a concept, what sends it back for a fix and what clears it.
At RIOT
Concept testing at RIOT
The Meta ad engine we ran for Homebase from February to April 2025 was concept testing at feed speed: creative directions put in front of real audiences with real money, read against what they actually did, and the weaker directions replaced. It is the live, in-market version of the storyboard in a focus group, and for performance work it is usually the better one.
Watch out
What people get wrong about concept testing
Letting the test write the creative. Research can tell you which concept is clearest; it cannot write the line or pick the cast, and a concept rebuilt around every comment in the readout ends up belonging to nobody. The other mistake is testing stimulus so rough that respondents react to the roughness. And the quiet one: testing to confirm a decision already made, then ignoring the result when it disagrees.
