Synthetic buyers vs real survey panels: what each is good for
AI-simulated buyers are fast and cheap; real panels are real. What the research and our own tests show, and which to use for which question.
"Synthetic" buyers are AI models asked to answer as a specific kind of person would: a 34-year-old parent in Ohio with a set budget, for example. Real panels pay real people to answer. Both are useful, and they are good at different questions. AskThousand gives you both: simulated buyers for a fast read on your page, and a free A/B poll where real people vote.
What the research says
A 2025 study by Maier and colleagues asked AI personas the purchase-intent questions from 57 real personal-care product surveys (9,300 human responses). Using a method where each persona answers in its own words and the answer is matched to a five-point scale, the AI reached about 90% of the consistency real people show with themselves when asked twice. It captured some group patterns well, such as age and income, but not others, such as gender and region.
Which to use for which question
| Your question | Best tool | On AskThousand |
|---|---|---|
| Why might people not buy this? | Simulated buyers: fast, many reasons | Free test, under a minute |
| Does my price look high next to competitors? | Simulated buyers, with competitor prices | Free test |
| What is missing from my description? | Simulated buyers | Free test |
| Which of these photos will people pick? | Real people's votes | Free A/B poll |
| What does a narrow niche think? | Real people from that niche | Share a free A/B poll in that community |
| How many units will I sell? | A small real launch | Test first, then launch small |
Strengths and limits
Synthetic buyers
- Answers in under a minute, so you can test many versions.
- Free or very cheap per test.
- Good at reasons, objections and missing information.
- Gives estimates, not facts, so use it to decide what to fix and what to test with real people.
Real panels
- Real people's choices and words.
- Best for visual decisions and narrow audiences.
- Paid panels charge per answer and take longer; panel members are not always your buyers either. A poll shared with your own audience avoids both.
A sensible way to combine them
- Use synthetic buyers to fix the obvious. Price, title, description, missing details. Re-run until the objections are minor.
- Use real people for the final visual choice. Main image, packaging, logo.
- Use real sales for the final answer. Launch small and measure conversion.
Questions
Are synthetic survey respondents accurate?
For purchase intent on product concepts, research has found AI personas close to real people on average, with weaker results for some subgroups. Use them as a fast first read, then let real people make the visual calls.
What is a synthetic panel?
A set of AI-simulated respondents, each given a profile such as age, income, region and habits, who answer the same questions a real panel would.
Is a real panel always better?
Not always. Real panels cost money per answer and take longer. For most sellers, simulated buyers for the page plus a free real-people poll for the visuals covers the same ground.
Check your product page in under a minute.
100 simulated buyers matched to your product's audience read your title, price and description and tell you who would buy, who would pass and why. Free, no account.