The cheapest moment to kill a bad idea is before anyone books a camera. The second cheapest is before anyone writes a script. After that the cost curve gets steep fast, and — more importantly — people start defending the idea because they've spent something on it.
This is a twenty-minute exercise. Seven questions, asked in order, most of them uncomfortable.
1 · What is the one sentence?
Write the idea as a single sentence a viewer would use to describe it to someone else.
Not the pitch. Not the strategic rationale. The sentence a person says when they pass it on.
If it takes two sentences, the idea has two ideas in it and needs splitting. If the only sentence you can write is a topic ("a video about our onboarding"), you have a subject, not an idea — and subjects don't travel.
2 · Who is this for, and what do they already believe?
Name the viewer specifically enough that you'd recognise them. Then write the sentence they'd say about this subject before watching.
That second part is the useful half. Content works by moving someone from one position to another — from not knowing, from being sceptical, from believing something that's wrong. If you can't state the starting position, you can't tell whether the piece moves anyone.
The most common failure here: the starting belief you write down turns out to be "I already agree with this." Content that confirms what the viewer already thinks feels good to make and does very little.
3 · Why would anyone watch past ten seconds?
Not "why is this valuable" — value is a reason to have watched. This is about the first ten seconds, when nobody knows it's valuable yet.
Answer concretely: what is on screen, what is being said, and what tension or promise does it establish? See how hooks work for the structural version.
If the honest answer is "because it's about a topic they care about", that's weak. Topic interest gets you a click. It doesn't survive a slow open.
4 · What's the evidence this format works — for us?
Three tiers of answer, in descending order of strength:
We've made things like this and here's how they did. Strongest. First-party, and it knows your audience.
Competitors make this and here's how theirs do. Good. Public tier 1 evidence, but it's their audience and their presenter, not yours.
It works in other categories. Weakest, and still worth something — but treat it as a hypothesis to test cheaply, not a reason to commit.
If the answer is "it feels right", you've located the actual state of the evidence. That's not necessarily fatal — some things are worth making on conviction — but it should change the size of the bet, which is the whole point of knowing.
5 · What would have to be true for this to fail?
Invert. Argue against it properly, as though you were being paid to.
Common ways ideas fail, worth checking each:
- The premise is interesting to us and not to the audience
- It requires context the viewer doesn't have
- Someone in the category made it already and better
- The good version needs a budget we don't have, and the affordable version isn't worth making
- It needs someone on camera who won't be good on camera
- The interesting part is legally or commercially unsayable
That last one kills more ideas at the script stage than any other, and it's knowable on day one.
6 · What's the cheapest version that would tell us?
If the idea survives to here, the next question is not "how do we make this well" but "how do we find out whether it's worth making well."
Almost every content idea has a cheap version: a talking-head take instead of the produced piece, a written post instead of a video, one segment instead of a series. The cheap version answers the demand question without answering the craft question, which is usually the order you want.
The exception is ideas whose entire value is production ambition. A piece whose point is that it looks expensive can't be tested cheaply — and that's worth knowing explicitly, because it means you're taking the full bet or none of it.
7 · What will we do with the result?
Decide now, in writing, what a good result and a bad result would each cause you to do next.
If a strong result wouldn't change anything — if you were going to make the follow-up series regardless — then the piece isn't testing anything and you should stop pretending it is. If a weak result wouldn't stop you, same.
This is the discipline from running a content experiment, applied one level earlier.
What to do with the answers
Strong on 1, 2, 3 and 5? Make it. Even if the evidence at question 4 is thin, an idea with a clear sentence, a named viewer, a real hook and survivable failure modes is worth a cheap version.
Weak on 1? Not ready. Rewrite the sentence until it works, or drop it. This is the most common outcome and the most valuable one.
Weak on 3? Usually fixable. The idea might be fine and the opening lazy — most ideas fail here because nobody's thought about the first ten seconds yet.
Weak on 2? Dangerous. This is the one that produces technically well-made content that lands on nobody, and it's the hardest to see from inside.
Weak on 5 — you can't think of a way it fails? You haven't tried. Every idea has failure modes. Not seeing them means you're too close, not that the idea is bulletproof.
Where Acumin fits
A Concept Read runs the evidence half of question 4 against your own history and your category's public content, and returns a scored read with a confidence band per dimension. On a genuinely novel idea the band comes back wide — which is the honest answer, and it's telling you the size of the bet rather than the answer.
What it can't do is questions 1, 2, 5 and 7. Those are judgement, and they're the ones that kill the most ideas. The intake conversation pushes on some of them, but a tool cannot tell you whether your premise is interesting to somebody who isn't you.
Run the seven questions first. Use the read to size the bet, not to make the decision.
How to use this today
Take the idea currently closest to production. Write question 1's sentence on a card and hand it to someone who wasn't in the meeting.
Ask them what they'd expect the piece to be. If their answer doesn't match yours, you've found the problem while it's still free.
Related: How hooks work, structurally is question 3 in depth. Running a content experiment is what to do when the honest answer to question 4 is "we don't know."