You have a new product idea, two or three versions of it, and a few weeks to find out which one deserves the development budget. The concept testing questions you write, and the order you ask them in, decide whether that answer can be trusted. This guide gives you a complete questionnaire in order, with example wording and the reason each question is there. For the wider method, including how to write the concept board itself, start with our guide to concept testing.
Key takeaways
- A concept testing survey runs screener, exposure, appeal and purchase intent first, then diagnostics, then demographics.
- Ask purchase intent straight after exposure, before any diagnostic question can make respondents think harder than a shopper would.
- Keep every question, scale and answer order identical across cells and concepts, or the comparison breaks.
- Open ends earn their place only when they are probed: ask what would change the answer, not just “any comments?”
- Check the gap between two concepts against the margin of error before you call a winner.
What concept testing questions should you ask, and in what order?
Each block in a concept testing survey has one job. The first half measures the overall reaction; the second half explains it. Adapt the wording to your category.
| Order | Block | Example question wording | Answer format |
|---|---|---|---|
| 1 | Screener | ”Which of these have you bought in the past 3 months?” plus an industry-employment check | Multi-select; exclude non-buyers and industry workers |
| 2 | Concept exposure | ”Please read the following description carefully. You will be asked about it afterward.” | Concept board, minimum view time |
| 3 | Overall appeal | ”Overall, how appealing is this product to you?“ | 5-point scale, not at all to extremely appealing |
| 4 | Purchase intent | ”If this product were available where you shop, how likely would you be to buy it?“ | 5-point: definitely would, probably would, might or might not, probably would not, definitely would not |
| 5 | Uniqueness | ”How new and different is this product from others available today?“ | 5-point scale |
| 6 | Relevance and need | ”How well does this product meet a need you have?“ | 5-point scale |
| 7 | Believability | ”How believable are the claims made about this product?“ | 5-point scale |
| 8 | Value for money | ”At $4.99 for a 12-oz pack, how would you rate its value for money?“ | 5-point, very poor to very good |
| 9 | Likes and dislikes | ”What, if anything, do you like about this product?” then “What, if anything, do you dislike or find confusing?” | Open text, one box each |
| 10 | Attribute agreement grid | ”How much do you agree or disagree that this product…” with 8 to 12 statements | 5-point agree scale, rows rotated |
| 11 | Comparison to current product | ”Compared with the [category] product you use most often, is this product…” | Much better to much worse |
| 12 | Frequency and replacement | ”How often would you expect to buy it?” then “Would it replace something you buy now, or be in addition?” | Single choice; replace, add or neither |
| 13 | Demographics | Age, gender, region, income | Single choice |
Let respondents view the concept again during blocks 5 to 12, so diagnostic answers rest on what it says, not on memory. Do not skip frequency and replacement: an idea people would buy once, or that would only replace your existing product, is a weaker case than its intent score suggests.
Why does purchase intent come right after exposure?
Purchase intent is the measure the decision usually rests on, so it should capture a reaction close to a real shopper’s, which is quick. Once you ask whether the claims are believable or the price is fair, respondents start evaluating analytically, and later intent answers reflect that extra thought.
So ask appeal and purchase intent immediately after exposure. Diagnostics then explain the intent score rather than shape it. Keep the standard five-point labels so you can compare top-two-box scores, the share choosing “definitely” or “probably would buy,” across studies. If you want to see where intent sits among awareness, consideration and trial, our purchase funnel guide maps the stages.
Why must the wording stay identical across cells and concepts?
In most concept tests, each concept goes to its own matched group of respondents, called a cell. Only the concept should differ between cells. Change a scale, reorder the grid or tweak the intent wording in one cell, and you can no longer tell whether a gap comes from the idea or from the concept testing questions themselves.
- Write one master questionnaire, with variables for the concept name and price.
- Use the same scales, anchors and rotation rules in every cell.
- Field all cells in the same period, matched on screener quotas.
Whether each respondent sees one concept or several changes the rules; monadic vs sequential monadic designs explains the trade-off.
How do you write open-end probes that diagnose?
“Any other comments?” produces short, vague answers. In our experience, the most useful open ends tie the probe to a closed answer the respondent just gave.
- Ask likes before dislikes, to avoid a critical frame.
- Follow a low intent score with “What is the main reason you would not buy it?”
- Follow a low believability score with “Which part is hard to believe, and why?”
- Ask everyone “What one change would make you more likely to buy it?”
Code answers into themes before pulling quotes. If one claim keeps coming up as hard to believe, test reworded versions; message testing covers how to compare claims head to head.
How long should a concept testing questionnaire be?
The flow above is around 15 to 20 questions plus the grid, enough to explain the score without wearing respondents out. Tired respondents straight-line grids and give one-word open ends.
Cut in this order: extra grid statements, secondary open ends, optional profiling. Never cut appeal, intent or value, because they carry the decision. Check median completion time and drop-out points in a soft launch.
What does a worked example look like?
The numbers below are illustrative, not from a real study. Suppose you test two versions of a snack concept, each shown monadically to 200 category buyers, and purchase intent comes back like this.
| Purchase intent | Concept A | Concept B |
|---|---|---|
| Definitely would buy | 18% | 14% |
| Probably would buy | 34% | 31% |
| Might or might not | 28% | 30% |
| Probably would not | 12% | 15% |
| Definitely would not | 8% | 10% |
| Top-two-box | 18 + 34 = 52% | 14 + 31 = 45% |
Concept A leads by 7 points. Is that a real win? The standard error of the difference is the square root of (0.52 × 0.48 / 200 + 0.45 × 0.55 / 200), which is the square root of 0.0025, or about 5.0 points. Multiply by 1.96 and the 95% margin on the difference is about 9.8 points. A 7-point gap sits inside it, so the honest read is “A is ahead, but not reliably.” Now the diagnostics do the work: if A scores clearly higher on uniqueness and the likes point to the same feature, that supports A. If not, you need a larger sample.
What are the common mistakes?
- Asking diagnostics before intent. Price and believability questions make people more critical, distorting intent.
- Changing wording between cells. A small edit can create a gap that looks like a concept difference.
- Testing without a price. Intent runs high when cost is invisible. Show the planned price, or ask value at a stated price.
- Leaving open ends unprobed. Tie each probe to a closed answer.
- Calling a winner on a small gap. At 200 per cell, a 7-point difference in top-two-box is not significant at 95%. Check the margin before you report.
How do you put the results to work?
Once fieldwork closes, read intent by the segments that matter, such as current users versus non-users. Crosstabs and banner tables with weighting and significance testing let you see which gaps between concepts and segments hold up. As you run more tests, connecting findings across studies and waves helps your team reuse what earlier concepts taught instead of starting over.
For your next step, copy these concept testing questions into your survey tool, fill in your category, price and concept names, and soft-launch with a few dozen respondents to check timing and open-end quality.