Concept optimisation

Concept optimisation is the research step that improves a concept after it has been tested. Most often it is a product concept, sometimes a marketing or advertising one. It finds which element of the concept holds it back, rewrites that element, and retests the new version against the original on the same audience, until further changes stop improving the result.

A concept test judges ideas as they are. Optimisation takes the most promising one and makes it better. It is where upstream research earns its keep: the difference between a concept that tests well and one that tests best is usually a sentence, not a new product.

The essentials

• Optimisation is a loop: test, diagnose, rewrite, retest. Each version changes one thing and says what it changes.
• The diagnosis decides what to rewrite. Without it, variants are guesses.
• Synthetic respondents make the loop affordable: a round takes hours, so it can run several times before fieldwork.
• The final version still needs confirmation with real people.

How does concept optimisation work?

1. Test. Measure the concept on the usual battery: comprehension, relevance, credibility, uniqueness, appeal, intent.
2. Diagnose. Find the element that costs the most. It is for a product concept often the headline benefit, the reason to believe, the formula, the format, the pack, the price, or a word the audience reads differently from the brand. The open-ended answers usually name it.
3. Rewrite. Write a few versions that each change that element, and record what each one changes. A version that changes three things at once teaches nothing, even if it wins.
4. Retest. Test the new versions against the original, on the same audience and the same questions. Keep what moved the result, drop what did not.
5. Stop. When the gain between rounds becomes smaller than the noise between two runs, the concept is ready for a confirmation test.

What is the difference between concept testing and concept optimisation?

Concept testing ranks ideas as they are. Concept optimisation improves the one that is worth it. Industry glossaries also use the term for studies that measure how much each benefit or feature contributes to a concept’s appeal; that is the diagnostic half of the same loop.

What do synthetic respondents change?

The cost of a round. In conventional research, a retest means a new sample and a new wait, so most concepts are optimised once, if at all. With a calibrated synthetic survey, the same audience can be asked again the same afternoon, and the loop can run until the gain stops rather than until the budget does.

What are the traps in synthetic concept optimisation?

• Optimising for the panel, not the market. A concept rewritten many times against the same simulated audience can learn what that audience likes. Confirm the final version on real people.
• Gaps too small to trust. Optimisation is read on differences between versions. If simulated respondents agree with each other more than real people do, those differences shrink or swing. We measured this under-dispersion in eight markets: 0.20 to 0.76 of the human spread before correction, 0.96 to 1.07 on the attitudinal layer and 0.85 to 1.00 on the personality layer after it.
• Winning on the wrong measure. A version can gain appeal and lose credibility. Read the whole battery, not the top line.

What should a buyer check?

• That every version states what it changed from the original.
• That versions were retested against the original, on the same audience and questions.
• That the winning gap is larger than the gap between two runs of the same version.
• That the final version was confirmed with real respondents.