Concept testing
A concept test shows people an idea before it is made or launched, such as a product, an ad, a pack or a name, and measures how they react: how much they like it, what they remember, and whether they would buy it.
The Ad or concept test template starts one. Compare two ads or concepts. Each person sees one of them.
Three designs
| Design | What each person sees | Trade-off | How to build it |
|---|---|---|---|
| Monadic | One concept | No concept affects how people see the next, but each concept needs a group of its own | A group for each concept, as in the template |
| Sequential monadic | Each concept, one at a time, rating each before the next | Fewer people, but each concept can affect how people see the next, so the order must be balanced | The concepts as parts of a section in a counterbalanced order |
| Side by side | Two or three concepts side by side, choosing one | A direct choice, which says which concept wins but not by how much | A preference test question |
The questions
The template asks these after each concept:
| Question | Type | Answers |
|---|---|---|
| What do you remember about what you just saw? Write anything that comes to mind. | Long text, not required | Anything they write |
| How much do you like this? | Rating, 7 points | Not at all to Very much |
| How likely are you to buy the product shown? | Rating, 5 points | Definitely would not buy to Definitely would buy |
Add your own on the same points for every concept, such as how new, how believable or how right for the brand each one is. Ask about the concept, not about the person: "How much do you like this?", not "Are you the kind of person who buys this?"
How many people each concept needs
For a monadic test, to find a difference of d between two concepts with 80% power and a two-sided test at 0.05:
| Difference (Cohen's d) | People for each concept |
|---|---|
| 0.2 | 394 |
| 0.3 | 176 |
| 0.5 | 64 |
A sequential monadic test needs fewer, since each person is compared with themselves.
Work out the people for your testReading the results
- Results shows how many people chose each point of each rating, and their share. Filter it by group to see each concept's answers.
- The top-two box is the share who chose the two highest points of the purchase rating: Definitely would buy and the point below it.
- For how far apart two concepts are, take each group's mean and standard deviation (SD) of a rating from the download, and work out Cohen's d.
What people say they will buy predicts what they buy better for existing products than for new ones, over a short time than a long one, for a named brand than for a whole category, and when concepts are compared than when each is rated alone (Morwitz, Steckel and Gupta, 2007).
Work out Cohen's dThe Ad or concept test template
- Instructions, then one empty image or video part for each of two balanced groups. Each is followed by a recall question, an attitude rating and a purchase rating. The face camera is on.
- You add an image or a video for each concept.
- The camera records the face only. Its video never leaves the computer that runs the session, and only the face measures are saved.
Source
Morwitz, V. G., Steckel, J. H., & Gupta, A. (2007). When do purchase intentions predict sales? International Journal of Forecasting, 23(3), 347–364.
Related
Templates
Start a study from a template, such as a survey, a pack test or a tasting.
Click and preference tests
Clicks on a picture, and a choice between pictures side by side.
Counterbalancing
Balanced orders of conditions, with a Latin square generator.
Face and gaze
Facial expression, blinks, head movement and gaze from a standard webcam.
Planning a study?
Start one now, or tell us what you want to measure.