How do you know if a startup idea is any good?

When somebody pays before you have built it. Every other signal — encouragement, survey results, sign-ups — is people being polite at no cost to themselves.

In more detail

The cheapest version of that test is a page describing the thing and a way to pay for it. What you learn is not whether people like the idea but whether they will part with money, and only the second one predicts anything.

What to do about it

An idea is worth almost nothing on its own. What is scarce is somebody who will keep going after the first version does not work, which is every first version.

The best ideas usually come from a job somebody already had. You notice a problem because you lived inside it, not because you brainstormed.

A market that already spends money is easier than a market you have to convince exists. Competition is evidence, not a warning.

The mobile app angle

If the idea is a shop, the sequence matters: sell first, prove people come back, then build the app. An app before the customers is a cost with nothing to do.