Say the claim out loud, as one sentence, to someone whose reaction you trust — then notice what you had to add before it made sense to them. If it needed three exceptions, or nobody could tell you what would prove it wrong, it is not ready for real time yet. The cheapest test costs an hour; the mistake it prevents costs a week.
Flow diagram of 4 steps: State the claim in one breath, Watch what you reach for next, Ask what would change their mind, Notice a clarifying question you cannot answer in one sentence.
The reason to test an idea before writing, recording or pitching it is not caution for its own sake — it is that the two activities have wildly different costs. Saying a claim out loud to one person and watching their face takes five minutes. Writing eight hundred words around that same claim, or recording and editing a segment built on it, takes a day or more, and by the time it is finished you are far less willing to throw it away than you were at minute five. That reluctance has a name — sunk cost — and it is exactly why the test has to happen before the investment, not after. Once a week of work exists, judging the underlying idea on its own merits gets harder, not easier.
This is not an argument for testing every idea exhaustively before starting anything. Small, cheap ideas — a single paragraph, a five-minute segment — can often just be written and judged on the finished result, because the cost of being wrong is already low. Testing earns its place specifically when the commitment ahead is large enough that being wrong would actually hurt: a multi-part series, a pitch to an editor, a segment built around a guest’s time as well as your own.
The fastest test is also the oldest one: say the idea to a person who will react honestly, using no more than one sentence, and pay attention not to whether they like it but to what you had to add to make it land.
No warm-up, no context paragraph — the actual claim, in the same words you would use if you only had one sentence to spend.
If you immediately add ’but only in this specific case’ or ’well, except when—’, that qualification is the real, narrower idea. Write that one down instead.
If they can name a case that would disprove the claim and you cannot answer it, the claim is not tested yet — it is just unchallenged.
That gap is usually exactly where the idea is still vague, not where the listener misunderstood you.
Every claim worth testing has at least one case that seems to break it, and finding that case yourself is far cheaper than having a reader, commenter or editor find it after publication. The method is deliberate, not passive: think of the group, situation or timeframe least likely to fit your claim, and check what actually happens to the claim there, rather than waiting to be told.
Take a claim like ’a shared task builds trust faster than a shared conversation’. The obvious exception to hunt for is a shared task nobody chose — a mandatory group project, an assigned chore rota — where the task is identical but the trust rarely follows. Finding that case does not necessarily kill the idea; it usually narrows it. The claim survives as ’a shared task builds trust faster than conversation, when the task is chosen rather than assigned’, which is a smaller, sturdier claim than the one you started with.
Where a conversation is not available, or the idea has already survived one, the next cheapest test is a miniature version of the real thing — small enough to abandon without regret if it does not hold up.
Full commitment versus a cheap test first, by format
| If the idea is for… | Full commitment | Cheap test first |
|---|---|---|
| A written guide | A finished 800+ word piece | The answer paragraph and one section, shown to one reader |
| A talk or recorded segment | A rehearsed, edited recording | A 90-second unscripted version, said out loud once |
| A pitch to an editor or producer | A full outline and draft | The one-sentence premise, sent alone, before anything else is written |
An idea has earned the time it is about to cost when it survives the one-sentence test without needing qualifiers added on the spot, when the exception you deliberately hunted for narrows the claim rather than disproving it outright, and when you can state plainly what evidence would prove the claim wrong. An idea has not earned that time when the sentence needed two or three exceptions before a listener stopped objecting, when the exception you found breaks the claim entirely rather than just narrowing it, or when you cannot say what would prove it wrong at all — a claim nothing could disprove is not yet a claim, whatever else it might be.
None of this requires more than one honest conversation and about fifteen minutes of your own attention. If a claim survives being said in one sentence to a candid listener, and survives you deliberately hunting for its worst case, it has earned the week it is about to cost. If it does not survive either, that is the fifteen minutes doing exactly its job — not wasted time, but time that would otherwise have been spent building something the same test would have flattened for free.
Usually under twenty minutes: one sentence said out loud to one honest listener, plus a few minutes spent deliberately hunting for the case that would disprove it. If the testing itself starts taking hours, it has quietly turned into the writing it was meant to precede — stop and write the small version instead.
That is a different problem from the one testing solves. Testing checks whether a claim is specific and defensible, not whether it is interesting. A claim can be precise and still be something few people want to read — the fix there is choosing a different claim, not testing this one harder.
No. Testing happens before anything is written, on the claim alone; feedback happens after a draft exists, on the execution. An idea that fails a test saves you from writing a draft at all. Feedback on a draft built on an untested idea usually just identifies the same problem later, at a higher cost.
Write the claim as one sentence and answer, in writing, what a specific sceptical person you actually know would say back to it — not a generic imagined reader. It is a weaker test than a real conversation, because it is harder to surprise yourself with your own objections, but it still catches a claim that needs several qualifiers before it makes sense.