Guide · 8 min read

How to evaluate a startup idea before you build it

A practical framework for judging a startup idea on commercial signals rather than enthusiasm: what to look at, in what order, and when to stop.

Most idea evaluation fails in the same way: it asks whether the idea is interesting instead of whether it is commercially survivable. Interesting is easy. Almost every idea sounds plausible once you have said it out loud twice. What follows is a sequence for judging an idea on the parts that decide its fate, and for knowing when the honest answer is to stop.

Start with the buyer, not the product

The first question is not what the product does. It is who has this problem badly enough to change their week over it. If you cannot describe that person in one sentence (their role, their situation, what they currently do instead), you are evaluating a category, not an idea.

Write the substitute down. Every idea competes with something, and usually that something is a spreadsheet, a document, or simply tolerating the problem. A free substitute that mostly works is the hardest competitor in any market.

Separate demand from willingness to pay

These get collapsed constantly, and they behave differently. Demand is whether people are actively looking. Willingness to pay is whether the problem is expensive enough to open a wallet. Plenty of ideas have real demand and no budget attached to it.

Test them separately. People searching for a solution tells you the first. People already paying for something adjacent tells you the second.

Judge the business shape, not just the product

An idea can be genuinely useful and still be a bad business. Four things decide the shape:

Recurrence: is this bought again, or once? A one-off purchase needs constant new acquisition.
Margin: what is left after the cost of delivering it? Per-use costs that scale with heavy users quietly eat the model.
Competition: how crowded is it, and what would stop the incumbent from adding your feature next quarter?
Scalability: can it grow without you personally being in every delivery?

Then ask the question most founders skip

Can you actually reach this buyer, repeatedly, at a cost the price supports? Distribution is where most promising ideas die, and it is almost always the last thing examined. If your channel plan is 'content and word of mouth' in a category saturated with content, you do not have a channel plan yet.

A product problem is a delay. A distribution problem is a wall.

Write down the assumptions carrying the idea

Every idea rests on a small number of load-bearing beliefs. Name them explicitly: the two or three statements that, if false, make the whole thing collapse. This is the single highest-value step in any evaluation, because it converts a vague feeling of risk into a testable list.

Then rank them by how cheaply they can be tested. The order you test in matters more than the number of tests you run.

Decide, and let the decision be one of three

An evaluation that ends in a score has not finished. Force it into an action: move forward, test one thing first, or stop. Two of those three save you time immediately, which is why an honest evaluation is worth more than an encouraging one.

Keep the framework identical between ideas

The value compounds only if the second idea is judged the same way as the first. Ad-hoc evaluation produces scores that cannot be compared, which means your portfolio never tells you anything. Same signals, same scale, every time.