Choose what to build

The best ideas are boring to everyone except the people who have lived them. They come from a job where you watched the same thing go wrong every week, from a process you had to work around, or from a market you know well enough to know what nobody says out loud.

A trends list gives you the opposite: problems you have never seen, in markets you cannot get a meeting in, with competitors you do not understand. You will spend your first six months learning what an insider already knows.

So start from your own record. Write down the work you have actually seen at close range. Not what you find interesting, what you have watched happen. The unfair advantage is that you know which parts are painful and which parts only look painful from outside.

A business removes a cost that someone is paying right now. Hours, headcount, error rates, refunds, churn, cash tied up. If you cannot name the cost, name who pays it, and say how it is measured, you have an interest, not a business.

Be specific about the unit. 'Saves time' is not a cost. 'Forty minutes per disputed handover, about 90 disputes a month, handled by an agent on 8,000 dirhams' is a cost. The second version tells you what you can charge and who signs the invoice.

Do this for each candidate. The ones that survive are the ones where you can write the arithmetic on one line.

Assume the build is free. With current tools a competent person can put a working version of most software on a real URL inside a week. So the interesting question is not whether you can build it. It is what stops the next person doing the same thing in week two.

There are only a few honest answers: proprietary data that accumulates, distribution you already own, a regulated or credentialled position, deep workflow integration that is painful to rip out, or a brand and trust position in a narrow market. Being first is not one of them. Nor is being cheaper.

Write your answer plainly. If the honest answer is 'nothing', that is useful information, not a failure. It usually means you should pick a narrower market where trust and service are the moat, or move to the next candidate.

Founders compare ideas on enthusiasm, which is why the newest idea always wins. Score them instead, on the same criteria, on the same day.

Six criteria, one to five each: how close you are to the problem, size of the cost removed, ease of reaching the buyer, how defensible it is in a year, how fast you can prove demand, and whether you would still want to work on it in three years.

The score is not the decision. It is a way of making your reasoning visible so you can argue with it later, and so a second reader can see what you weighted.

Write down the one you are building, the date, and the reason. Not because the decision is irreversible, but because a written reason is the only way to tell later whether you changed your mind on evidence or on mood.

Also write your kill condition now, while you are calm: the specific thing that would make you stop. It is much cheaper to define that today than in week nine.