Will people buy your new product? Ask a better question.

A launch test needs a price, alternatives, and a reason to switch.

'Do you like this idea?' is an easy question to answer. 'Would you buy it instead of what you use now, at this price?' asks something much closer to the decision a customer will face.

Imagine testing a new subscription. Show the monthly price, what it includes, and the service a customer would give up. Include keeping their current plan as a choice. Otherwise, you might learn that the idea sounds appealing without learning whether anyone would switch.

Ask about timing too. Someone might want the product but have no reason to change today. A contract, a setup cost, or the effort of learning something new could affect the decision.

Use an early test to decide what deserves more work. Then compare the result with a small real-world trial when possible. Keep the questions consistent across human and AI research so you know which choice each result describes.