Idea APicture + text
Brand A
Chili lime crunchTortilla chips with a squeeze of lime and a slow, warm chili.
Concept testing
Find the winning idea before you build it
Show your new ideas to the people you want to buy them, and learn which one to take forward and why. Each person sees one idea and answers the same questions, then tells you in their own words what works and what does not.
Idea BPicture + text
Brand A
Sea salt & cracked pepperThick-cut chips, sea salt and coarse black pepper.
Idea CPicture + text
Brand A
Ghost pepper mangoSweet mango first, then the hottest pepper we could find.
Each person sees one idea · groups fill evenly by age and gender
1idea per person, so no idea sways the next
100the top of every score, which runs from 0
4plain verdicts, each with its reason
200people per idea we suggest, at least
Why test it first
Check the idea while it is still cheap to change
Many new products struggle not because they were made badly, but because nobody checked the idea with buyers first. Once production, the packaging and the media plan are booked, every change costs more. A concept test answers four questions while change is still cheap. Which of our ideas should we take forward? Would people buy it, and does the price put them off? What do they like, and what holds them back? Is it different enough from what is already on the shelf?
What can go on the bench
TextA pictureA picture with textA video that plays to the endA productA serviceA menu itemAn offerA featureA new way to position a product
One test, start to finish
Watch a concept test run, start to finish
- 01The idea goes on the bench
- 02People react, in the same order
- 03Scores build up, from 0 to 100
- 04The verdict, with its reason
Step 1 / 4
The idea goes on the bench
Every idea uses the same format, so the format never picks the winner. Each person sees one idea only.
Study set-up
Concept testing
- StimulusThree new flavors, each a picture with a short description
- WhoUS salty-snack buyers who bought the category recently
- DesignEach person sees one flavor; groups fill evenly by age and gender
- QuestionsTrial first, then feeling, appeal, newness, relevance, clarity and believability
- PriceShown after the core questions, so the first answers stay clean
- People300 per flavor, 900 in all
Set up with our team, or yourself in CatalyseAI
Step 2 / 4
People react, in the same order
Would they buy it? Which face is closest to how they feel? Then likes and dislikes in their own words, and the price last.
Question 1 of 14
How likely would you be to buy it?
- Definitely would
- Probably would
- Might or might not
- Probably would not
- Definitely would not
Which best describes how you feel about it?
LoveExcitedHappyCuriousSurprisedConfusedUnimpressedPut off
"The lime makes it. Warm, not hot."
"Tastes like the chips I already buy."
"Way too hot for me, honestly."
0:38
On-camera answer
AI follow-up: What would make you pick it up in the store?
"If the bag said how hot it is, I would try it straight away."
- Live person
- Light
- Length
- Relevance 0.86
- Attention 0.91
Transcribed, checked and searchable. Up to three follow-ups per answer.
Step 3 / 4
Scores build up, from 0 to 100
The top answer counts in full and the next answer counts half, so every score can be traced back.
900 people answered
Chili lime crunchSea salt & cracked pepperGhost pepper mango
Trial
- 64
- 61
- 38
Newness
- 62
- 34
- 71
Appeal
- 66
- 63
- 44
Step 4 / 4
The verdict, with its reason
Each idea gets one of four plain verdicts and one sentence with the numbers behind it.
The scorecard
What we measure, in plain names
Each row is one measure, named the way the report names it.
Base: 300 salty-snack buyers per flavor, and each saw one flavor only.
Scores run from 0 to 100. A small letter marks a score that is significantly higher than that idea's score.
Measure
AChili lime crunch
BSea salt & cracked pepper
CGhost pepper mango
TrialWould people buy it, or try it? Asked first, straight after the idea, so nothing else sways the answer.
64C
61C
38
FeelingThe share who pick a happy, excited or loving face when asked how the idea makes them feel. People choose the face themselves; we never read faces from video.
58C
49
41
AppealHow much people like the idea.
66C
63C
44
NewnessHow different it is from what they can buy today, and how much better than what they buy now. Both halves are always shown.
62B
34
71AB
RelevanceHow well it fits what they need or want.
60C
64C
39
ClarityHow easy it is to understand what the idea is.
72C
78AC
55
BelievabilityWhether people believe what the idea promises.
63C
70AC
48
Trial at price, Value and Price effectWith a price shown: would they still buy it, is it good value, and how much Trial moves once they see the price.
57CPrice effect -7
52CPrice effect -9
30Price effect -8
Likes and dislikesWhat appeals and what puts people off, in their own words.
"The lime makes it." "Warm, not hot."
"Tastes like the chips I buy now."
"Way too hot." "Love the mango."
VerdictJudged against each other
Promising
Wanted, not new
New, not wanted yet
The verdict
Example study: three new chip flavors
A US snack brand has three new tortilla chip flavors and room to launch one. It shows each flavor as a picture with a short description to 300 US salty-snack buyers per flavor, 900 people in all, and adds the price after the core questions. Each person sees one flavor. The first flavor comes out Promising: people want it and find it new. The second is Wanted, not new: people would buy it, but it feels like chips they already have, so the team rewrites what makes it different. The third stands out, but fewer people want it, and the dislikes say the heat is too strong. Trial holds up for the first flavor once the price is shown, so the team takes it forward and studies the price next.
The four verdicts
- Promising
- Wanted, not new
- New, not wanted yet
- Not ready
Chili lime crunch
Promising
People want it and find it new: Trial 64, Newness 62.
Check the price next: Trial holds at 57.
Sea salt & cracked pepper
Wanted, not new
People would buy it, but Newness is 34: it feels like chips they already have.
Rewrite what makes it different.
Ghost pepper mango
New, not wanted yet
It stands out, Newness 71, but Trial is 38 and the dislikes say the heat is too strong.
Tone the heat down and test again.
Ask the studyAnswers from your own data
Does Chili lime still lead once the price is shown?
Yes. Trial for Chili lime drops from 64 to 57 with the price, and it still leads. Ghost pepper mango falls furthest, to 30.
It checks its numbers against your data and draws the chart.
How it works
How a concept test works
From brief to decision, step by step.
01
Tell us what you are deciding
Share your brand, the category and one to five ideas. An idea can be a product, a service, a menu item, an offer, a feature or a new way to position a product you already sell. Our team can set it up with you, or you can set it up yourself in CatalyseAI.
02
Show each idea the way people will meet it
Every idea in a study uses the same format: text, a picture, a picture with text, or a video that must play to the end. Each person sees one idea only, so the ideas never color each other's scores. The idea groups fill evenly by age and gender, so they stay alike.
03
Ask the same core questions about every idea
Would they buy it? Which face is closest to how they feel about it? How much do they like it, and what do they like and dislike, in their own words? Then how different and how much better it is, how well it fits their needs, how clear it is and how believable. If you add a price, it comes after these questions, so the first answers are never colored by it.
04
Field it with buyers in your category
The study runs on PocketsFull, our own panel. A short screener lets in only people who bought the category recently. If you turn on the video question, people also answer on camera, and an AI interviewer can ask up to three follow-up questions.
05
Read the verdict, then the reasons
Each idea gets plain scores from 0 to 100 and a one-line verdict. You choose what the ideas are judged against: each other, targets you set, or your current product. Under the verdict sit the likes and dislikes, grouped into themes, with quotes.
Then: decide
Promising: wanted and new. The price study comes next.
Plan your study
What you get
What you get
Output 01
A verdict for each idea
One of four plain verdicts: Promising; Wanted, not new; New, not wanted yet; or Not ready. Each comes with one sentence that gives the reason and the numbers behind it, plus one thing to watch.
Output 02
Scores you can trace back
Every score runs from 0 to 100. The top answer counts in full and the next answer counts half, so you can always see how a score was reached.
Output 03
Ideas side by side
One table with every idea and every score. Letters mark the gaps that are statistically real, so a two-point difference is never sold to you as a win.
Output 04
Likes and dislikes in their own words
Open answers grouped into themes, with quotes, so each number comes with a reason you can read out in the meeting.
Output 05
Picture taps
For picture formats, people tap the parts they like and dislike. You see the heat on the picture and, if you name areas such as the logo or a claim, the share who tapped each one.
Output 06
Video answers, if you want them
Transcripts in any language, themes across all the answers, and each video checked for attention, relevance to the question and lighting.
Output 07
Files to share
A PowerPoint deck, an Excel workbook, CSV data and chart images.
FAQ
Questions, answered
How many people should see each idea?
We suggest at least 200 people per idea, and 300 if you want to compare groups such as age bands or regions. Each person sees one idea, so three ideas at 300 each means 900 people. Below 100 people per idea we still show the scores, but no verdict.
Do you compare my scores with other studies?
No. We do not have a database of past scores to compare with, so we do not pretend to. Each study is judged on its own: ideas against each other, against targets you set, or against your current product shown the same way.
Will a high score tell me how much it will sell?
No. A concept test compares ideas and shows what to fix. It is not a sales forecast, and the report says so.
When should I use another method instead?
If you have a long list of rough ideas, screen them first and test the shortlist. If you are choosing among many claims or features, MaxDiff ranks them more cleanly. If the idea is settled and the open question is the price, run a pricing study.
Which countries can I test in?
The ready-made concept test in CatalyseAI runs with US audiences first. For other countries, talk to our team and we will check what is possible for your audience.
Related
Goes well with
Choosing between ideas?
Tell us what you are testing and who you want to sell to. We will suggest the format, the sample size and what to judge the ideas against.
Nearest office: Orlando, FL (corporate office)