The 48-hour test that beats asking people if they'd use it

Enthusiasm is free, so it means nothing. Four validation tests that cost the other person something, ranked by how much signal they carry.

2026-06-16 · 4 min read

"Would you use something that did X?"

"Yeah, definitely, that sounds really useful."

That exchange has preceded more wasted months than any other in software. It is not that people are lying — they mean it at the moment they say it. It is that agreeing costs nothing, and a signal that costs nothing carries no information.

The fix is to design a test the other person has to pay something for. Not necessarily money. Anything scarce works: time, reputation, access, effort.

The ladder, weakest first

Level 0 — they say yes. Costs nothing. Worth nothing. Everybody starts here, and the mistake is stopping here because the answers are so encouraging.

Level 1 — they describe the last time it happened. "When did you last do this task? Walk me through it." This costs a few minutes and, crucially, it cannot be answered by imagining a future. If they cannot recall a specific instance in the last month, the problem is hypothetical. Roughly half of enthusiastic yeses die here, which is why it is the first question worth asking.

Level 2 — they hand over something. A real input file. Access to the folder. Twenty minutes on a call with the actual spreadsheet open. This costs effort and a little trust. People who will not send you a sample file will also not adopt the thing that processes it.

Level 3 — they commit their name. They agree to be the pilot, in writing, with a date. Or they introduce you to the person who owns the budget. Reputation is on the line now, which is why this is where most polite interest stops — and that stopping is the finding.

Level 4 — they pay. A deposit, a pre-order, a paid discovery session. Even £50 changes the conversation completely, because the person now has an interest in the thing existing rather than an opinion about it.

You do not need Level 4 to proceed. You do need to know which level you actually reached, and to stop reporting Level 0 results to yourself as though they were Level 3.

What 48 hours actually buys

The constraint is not arbitrary. Two days is long enough to run a real test and short enough that you have not built anything you would feel obliged to defend.

A workable two days:

Hours 1–4. Write down the single sentence you are testing, and — this is the part people skip — the result that would make you stop. "Fewer than three of ten say they did this in the last month" is a stopping rule. Without one written in advance, every outcome will look encouraging in hindsight.
Hours 5–20. Talk to ten people who have the problem. Not ten people who find it interesting. Ask the Level 1 question and shut up. The transcript matters more than the tally.
Hours 21–36. Ask the five most engaged for a Level 2 commitment: send me the file, show me the process. Count who actually does it within 24 hours, not who agrees to.
Hours 37–48. Write up what happened against the stopping rule you wrote on day one.

The output is not a decision to build. It is a page of evidence with a number on it: ten conversations, six had done the task in the last month, four sent a file unprompted, one asked what it would cost. That is a real result, and it took two days.

The failure mode of this method

Two, and they are worth naming because this technique gets oversold.

You can validate the wrong sentence. "People spend ages on invoices" validates easily and tells you nothing about whether they will change tools, which is a completely separate question with its own answer. Test the switch, not the pain.

Small samples lie in both directions. Ten conversations can miss a real market and can also manufacture a fake one if you found the ten most sympathetic people you know. Ask yourself who you did not talk to.

The point

You are not trying to prove the idea is good. You are trying to buy the cheapest possible answer to "is anyone doing something about this today", before the answer costs you a month.

Two days and a stopping rule written in advance is about as cheap as that gets. Validate it in 48 hours runs the full version — including how to read the answer when it is neither yes nor no, which is what it usually is.

Take it further

Validate it in 48 hours — Find out whether anybody wants it before you spend six weekends finding out they don't. (3 lessons, 65 min, 3 free)
Sell the thing, not the hours — Why your rate has a ceiling, where the repeatable part of your work is hiding, and what a product costs you after the sale. (3 lessons, 85 min, 1 free)