How to validate an idea

Validation is the deliberate search for evidence you are wrong, done cheaply, before you commit months of work.

The three questions, in order

1. Is the problem real and frequent? (discovery)
2. Will someone pay to remove it? (willingness to pay)
3. Can it be delivered reliably enough to trust? (feasibility)

Most ideas die at question 2, not question 3. Test money before you test technology.

Cheap tests that produce real signal

• The paid pilot letter of intent. A signed agreement to pay for a small first engagement is worth more than a hundred survey responses.
• The manual concierge. Deliver the outcome by hand for one customer, with no product at all. If it is valuable done manually, automation is an efficiency play; if it is not, no amount of AI will save it.
• The pre-sell. Offer the pilot at a real price before it exists. A "yes, here is my deposit" is validation; a "sounds interesting" is not.

Set the bar before you start

Decide the pass/fail line in advance, so you cannot rationalise a weak result later. For example: "Two of five operators must commit to a paid pilot within three weeks." Write it down. Honour it.

Distinguish the ladder of commitment

Not all "yes" is equal. In ascending order of signal:

1. "Interesting." (worthless)
2. "I'd use that." (weak)
3. "I'd pay for that." (better)
4. "Here is my calendar / deposit / signature." (real)

Only the bottom rung counts.

Record it honestly

Keep a note of what you predicted, what you tested, and what actually happened — including the times you were wrong. See evaluation and evidence for how to make results credible rather than anecdotal.

Next

If the idea survives, design a small pilot.