Starting a software business

Write the problem in one sentence without naming your solution. If you cannot, you have a product idea rather than a business idea.

Name the specific person who has this problem. Not a demographic — a person you could call.

Find where those people already gather and read what they complain about in their own words. Save the exact phrasing; it becomes your copy later.

Identify what they do today instead. Every problem already has a workaround, and that workaround is your real competitor.

Establish what they currently pay for that workaround — in money, time or hassle.

Talk to ten of them before building anything. Ask what they did last time the problem happened, not whether they would buy.

Write down what would have to be true for this to fail, then check whether any of it is already true.

Decide the smallest version that would prove someone will pay. Write it down and resist expanding it.

Confirm the problem recurs. One-time problems do not support subscriptions.

Identify whether you need data you do not have and cannot easily get. That is the most common silent killer.

Check whether anyone is already doing this well. A crowded market proves demand; an empty one often proves there is none.

Work out how someone would find you. If the answer is "word of mouth" you do not have a channel yet.

Set a decision date and a kill condition before you start. Without one you will keep spending on something out of stubbornness rather than evidence.

Ask whether you actually want to run this business day to day, not just whether it could work. Most founders skip this and regret it.

Estimate the smallest amount of money that gets you to a first sale, and check you can lose it without damage.

Check whether a spreadsheet already solves this. If it does, your product needs to be dramatically better, not marginally.

Decide whether you are selling to individuals or companies. It changes pricing, sales cycle and support entirely.