Rehearse Your Next Business Decision with HumanPanel

Raise prices by 10%? Open a second location? Add a delivery fee? Most business owners make calls like these on instinct, a few chats with regulars, and hope.

Big companies test decisions like these with research teams and survey budgets. Small businesses usually can’t justify that for every question.

HumanPanel gives you a cheap rehearsal. You describe your customers, ask them about the change, and see how a simulated panel of up to 1,000 people reacts, with a reason behind every answer. Most runs cost $1 to $2.

HumanPanel simulates how people might respond to a decision, not what your revenue will be. It is a rehearsal of customer reactions, not a financial forecast.

Here is what happens when you run one:

• You describe your customers in plain words, for example “regulars at a neighbourhood café in Leeds, mostly commuters and remote workers.”
• HumanPanel splits them into “worlds”, 1 to 10 segments such as morning commuters, students and weekend families. Each world is built from about 14 weighted attributes like age, income, education and media habits.
• Simulated respondents answer your questions. Each has a background, beliefs and blind spots, and explains every answer.
• You get a verdict broken down by world, with common themes, direct quotes and a written summary.

Two features matter for business decisions. Saved audiences let you test option A and option B on exactly the same simulated people. And you can chat with any respondent to ask “what would change your mind?”

Any decision that depends on how customers will react is a good candidate. Here are six common ones.

Keep each run focused on one decision. A run allows up to 20 questions, which is plenty for one change and its follow-ups.

Imagine Sam runs a small café and wants to raise coffee prices by 40p because milk and rent costs went up. Here is how Sam could set up the simulation. (This is an illustration of the setup, not real results.)

1. Describe the audience. “Regular customers of an independent café near a train station in a mid-sized UK city: commuters, remote workers, local students and weekend families.”
2. Check the worlds HumanPanel proposes. For example: weekday commuters, remote workers, students and weekend families. Sam can adjust these if they don’t match reality.
3. Write the questions. Six is enough:
   1. How often do you buy coffee here? (single choice)
   2. If your usual coffee cost 40p more, what would you most likely do? (single choice: keep buying as usual, buy less often, switch to a chain, make coffee at home)
   3. Would a loyalty card, such as every 8th coffee free, change your answer? (yes/no)
   4. Which would bother you least: a price rise, a smaller cup, or fewer pastry options? (single choice)
   5. Is this café better value than the chain nearby? (true/false)
   6. What, if anything, would make a higher price feel fair to you? (free text)
4. Run it, then run the alternative. Sam saves the audience and runs a second version with a 25p rise. Both versions face the same simulated people, so the difference comes from the price, not the panel.
5. Dig into surprises. If students react much more strongly than commuters, Sam can chat with a few student respondents and ask what would soften the blow.

Simulated answers are a directional signal, not a prediction of what real customers will do. Research backs up that caution.

Bisbee et al. (2024) found AI respondents matched real survey averages closely, but gave less varied answers than real people and shifted with small wording changes. Brand, Israeli and Ngwe found AI price estimates were sometimes close to human ones but often inaccurate.

So use the results to decide what to check next, not to skip checking:

• Trust the reasons more than the percentages. An objection you had not considered is valuable even if the exact numbers are off.
• Compare options, don’t predict totals. “Version A drew fewer objections than B” is a safer takeaway than “we’ll lose 12% of customers.”
• Confirm cheaply with real people. Put the winning option on a sign, ask ten regulars, or trial it for two weeks.

Pick one decision you are stuck on this month. Sign up at humanpanel.ai with your email, get $1 of free credit with no credit card, and write your survey for free.