Nine dimensions, assessed as evidence rather than as a score. What you know, what you assume, and what is still unvalidated.
15 min read · Level 1
An idea score is a way of avoiding the specific question you are afraid of. Collapsing nine dimensions into "82 out of 100" destroys the only information worth having: which parts you actually know and which you are guessing about. The useful output of evaluation is never a verdict. It is a ranked list of the assumptions that would hurt most if they were wrong, and a plan to test the top one this week.
Nine things determine whether an idea is worth years of your life:
Severity — how much does the problem hurt? Frequency — how often? Willingness to pay — will they actually part with money? Market size — how many have it? Competition — what exists, including doing nothing? Distribution — can you reach them affordably? Founder-market fit — why you? Feasibility — can it actually be built? Defensibility — what stops a copy?
Assess each as evidence, not as a score out of ten.
Because a single number — 'your idea scores 82/100' — is a false precision that hides the only useful information: which parts you actually know and which you are guessing.
The useful output is not a verdict. It is a list of the assumptions you have not tested, ranked by how badly you would be hurt if they were wrong.
A scheduling tool for small logistics firms:
Dimension — State — Evidence
Severity — 🟢 Strong — 14 of 15 interviewees called it their worst weekly task
Frequency — 🟢 Strong — Weekly, every week
Willingness to pay — 🟡 Unknown — Everyone said 'probably'. Nobody has paid.
Market size — 🟡 Needs work — Rough bottom-up estimate, unverified
Competition — 🟢 Understood — Two enterprise tools, too expensive; most use spreadsheets
Distribution — 🔴 Unvalidated — No idea how to reach these firms at scale
Founder-market fit — 🟢 Strong — Six years in the industry
Feasibility — 🟢 Strong — Straightforward to build
Defensibility — 🟡 Unknown — Nothing obvious yet
The two red and amber items — willingness to pay and distribution — are what to work on next. Not the product.
That is what this exercise is for. It does not tell you whether to proceed; it tells you what to do on Monday.
First-time founders often mark everything green because they believe in the idea. The exercise is only useful if you are willing to write 🔴 next to something you care about.
The second mistake: treating distribution as something to figure out later. It is the most commonly fatal red on this list — plenty of good products die because there was never an affordable way to reach the customer.
Do not skip the uncomfortable ones. Those are the ones with information in them.
Strong means you have evidence from outside your own head. Not a strong feeling — evidence.
'14 of 15 interviewees said X'. If you cannot cite something, it is not strong.
Which would hurt most if it turned out badly? That is the next thing to test, regardless of how hard it is.
Usually a conversation, a pre-sale or a landing page — not a build.
Before committing serious time, and again every few months as evidence accumulates.
Do not use this to decide whether to talk to customers. Talk to them first — most of these boxes can only be filled in with what they tell you.
Across post-mortems that founders publish after shutting down, an inability to reach customers affordably appears far more often than a bad product. Founders defer the distribution question because it feels like a later problem, and it is routinely the one that ends the company.
If you cannot name how you will reach your hundredth customer, that is your riskiest unknown — regardless of how good the product is.
The predictable failure of any self-assessment exercise: everything comes out green because the founder believes in the idea. The exercise only produces value at the moment someone is willing to write "unvalidated" next to something they care about.
If your evidence board has no reds, you have not done the exercise. You have described your hopes in a table.
Create a free account to read the rest and apply it to your own startup.