Guides
Use synthetic consumer research to narrow the next test
Simulated feedback can help screen a concrete consumer offer and price while changes are cheap. It does not replace recruited respondents or observed buying behavior.
Can simulated feedback narrow the next test?
Yes, when the job is early prioritization. WouldTheyBuy screens one concrete consumer offer and approximate price, then returns a directional buying-interest pattern, broad audience signals, reasons, and concerns. Use that output to decide what deserves stronger evidence, not as proof that people will buy.
Current product boundary
The public test accepts a product or offer description, approximate price, optional category, and optional differentiator. Include the intended consumer inside the description. The current panel uses broad U.S. demographic profiles across age, income, gender, region, and ethnicity; it is not a custom sample of your target market.
Each profile writes a reaction to the same offer. Those responses are summarized as a five-level buying-interest spread with reasons, concerns, and broad subgroup patterns. The profiles are simulated, and repeated runs can vary.
What the published evidence covers
WouldTheyBuy follows the approach evaluated in the arXiv preprint. The study covered 57 U.S. personal-care concept surveys with about 9,300 human responses. The simulated results placed concepts in a similar order to repeat human surveys, reaching roughly 90% of the agreement available between those repeat surveys. That is a directional ranking benchmark, not accuracy, purchase-prediction accuracy, or proof that the same result carries over to every category.
Limitations
- Stated interest is not a purchase. A simulation cannot reproduce real budgets, urgency, competing offers, checkout friction, or actual behavior.
- Subgroup patterns are not equally reliable. Some age and income patterns transferred better than gender, region, and ethnicity patterns in the paper.
- The benchmark is category-specific. The evaluated concepts were hypothetical U.S. personal-care products. Unfamiliar, niche, or highly specialized domains may produce weaker signals.
- The panel is broad, not custom. The product does not recruit or tightly screen your intended market.
- Hands-on and high-stakes decisions need more. Sensory experience, safety, claims, large inventory, investor, board, medical, and regulatory decisions require stronger evidence.
When it fits and when it does not
| Decision | WouldTheyBuy fit | Stronger evidence |
|---|---|---|
| Which written consumer concept or price deserves the next test? | Useful as a directional first screen | Human concept survey or interview |
| Which packaging, image, taste, fit, or physical experience wins? | Poor fit | Real respondents who can see or experience it |
| Will the offer convert in market? | Cannot answer | Presale, live ads, checkout, or purchase behavior |
| What will a specialist or B2B buying committee do? | Poor fit | Interviews and research with the actual roles |
The next stronger validation step
Change one element at a time, such as price, benefit, audience wording, or differentiator, and compare separate reports cautiously. Then move the clearer version into a product concept survey, a human-response test, or behavioral evidence. Synthetic consumer research earns its place by helping you ask a sharper question; the final decision still belongs to real people and real behavior.
Related guidance
Frequently asked questions
What is synthetic consumer research?+
It uses simulated consumer profiles to explore how people might react to a product, offer, or price. The responsible use is early prioritization, followed by research or behavior from real people.
Are the respondents real?+
No. Every response comes from a simulated profile. The current panel represents broad U.S. consumer demographics, not a recruited sample of your exact audience.
What did the published benchmark cover?+
The arXiv preprint evaluated 57 U.S. personal-care concept surveys containing about 9,300 human responses. It compared concept rankings with repeat human surveys; it did not measure sales-prediction accuracy.
When should I use real people instead?+
Use real respondents when the audience must be tightly screened, the product depends on hands-on or sensory experience, the decision is costly, or you need defensible evidence from actual people.
What should I test after the screen?+
Verify the most important uncertainty with customer interviews, a human panel, a concept survey, sensory work, a presale, live ads, or observed purchases.