Stimulus and Concept Preparation

Why does the same concept keep winning every test?

Check the writing before the idea. A concept with more words, a quantified benefit, a reason to believe its rivals lack, more evaluative adjectives, or an opening that frames a problem before offering the solution will outperform, and nothing in the data will say so. Audit the set in a column, dimension by dimension, using someone who did not write it. Expect the strongest-written concept to be the one its author already preferred.

How finished should test stimulus be?

Match fidelity to the decision, and keep it uniform across the set. Rough stimulus isolates the idea, is cheap enough to test many options, and invites respondents to imagine, which produces generous and unstable ratings. Finished stimulus is closer to what will exist and confounds idea with execution, because nobody can separate a good idea rendered badly from a bad idea. A set containing one finished item among roughs is not a test of ideas; it is a test of finish.

Should I use monadic or sequential monadic exposure?

Monadic, where each respondent sees one concept, gives the cleanest absolute read, is comparable to monadic norms and approximates how people actually encounter a proposition, at the cost of a full sample per cell and no within-respondent comparison. Sequential monadic is efficient and gives within-respondent comparison, at the cost of calibration against earlier items and fatigue in later ones. Comparative side-by-side is the most sensitive for close variants and the least realistic, because it creates a choice the market does not offer.

How many concepts can one respondent evaluate?

Three to five with full diagnostics is a common working range, fewer when items are long or complex, and it should be calibrated in pilot by looking at reading time and open-end length by position rather than taken as fixed. Where more concepts must be covered, use a balanced incomplete block design so each item appears equally often and equally often in each position.

Does adding a price and competitors lower concept scores?

Usually, and substantially. That is the point. Scores fall because the test moved closer to the world, and the differentiation between concepts generally improves at the same time. The real question is never whether someone likes a concept but whether it pulls them away from what they already do, and that question needs a competitive frame and a price.

What does a middling score with vague open ends mean?

Very often comprehension failure rather than rejection. A concept that is not understood scores in the middle, because respondents rate what they think it might be. This is why a comprehension pilot, in which people say back in their own words what is being offered and what is new about it, catches problems that ratings cannot.

Can a concept test predict sales?

No, and this is worth writing into the report before results exist. A concept test removes almost everything that determines real-world performance: availability, price in context, competitive response, media weight, repeat purchase, and the fact that nobody in the market will ever read the concept. It ranks options against each other reliably. Converting a concept score into a volume or share estimate requires a calibrated forecasting design and external data.

How do I test a genuinely new idea fairly?

Recognise the handicap. Familiar propositions are easier to write, easier to understand and score better, so a novel idea is systematically disadvantaged in short-exposure testing. Options include allowing more explanation and documenting the resulting parity disparity, testing it monadically against its own reference rather than in the set, extended or two-stage exposure, or qualitative evaluation before quantitative screening. Whichever is chosen, say in the report that a low score in the standard format is weak evidence for that class of idea.

Should concepts be branded or unbranded?

There is no universally correct answer, only a decision that must be made once and applied to every item. Branded stimulus measures the idea plus the brand's permission to make the claim. Unbranded stimulus measures the idea alone and overstates what an unknown entrant could achieve. What is never acceptable is one branded item in an otherwise unbranded set.

When should stimulus be changed mid-fieldwork?

Almost never, and never quietly. Changing stimulus while a study is live splits the sample into non-comparable groups. If a defect is found in field, the decision is whether to pause, and any change is a logged amendment that partitions the data into before and after. Version every item from the start, because late copy edits from outside the research team are how a report ends up describing something that was never fielded.