Creating ideas: where good ones actually come from

The short version. Almost nobody good at this waits to be struck by an idea. They keep a supply line: they collect problems, notice workarounds, take stock of what they can do that others can't, and generate lots of candidates cheaply — then let evidence, not affection, choose between them.

1. Start from problems, not from ideas

An idea invented in the abstract has to go looking for a problem afterwards, and usually doesn't find one. Reverse it. Keep a problem journal for two weeks: every time you, or anyone around you, mutters "why is this so hard", write it down — the errand that takes three apps, the form that gets printed and re-typed, the job everyone at work quietly dreads. Boring, repeated, mildly painful problems are the richest seam, precisely because nobody glamorous is looking at them.

2. Hunt workarounds — they are demand, already proven

Wherever people have built their own clumsy fix, someone has already voted with their time. The spreadsheet that runs an entire department. The WhatsApp group that functions as a booking system. The binder of printouts a tradesperson keeps in the van. A workaround is a customer telling you what they would buy, in the most credible language there is: effort they already spend.

3. Take stock of your unfair advantages

The same idea is weak from one person and strong from another. Before generating anything, write down honestly: what do you know that most people don't (a trade, an industry's plumbing, a regulation's real effect)? What do you have access to that others can't easily get (an audience, data, relationships, equipment, a licence)? What have you done repeatedly that others do once? The ideas worth your time sit at the crossing of a real problem and one of these — anyone can have the idea, but not anyone can do it.

4. Generate in volume, choose in cold blood

The first idea is rarely the best; it is just the first. Give yourself a rule — twenty candidates before you evaluate any — and use crossings to force variety: each problem from your journal × each advantage from your list. Most combinations will be nonsense. That is the method working, not failing: quantity first, judgement second, and never during.

5. Shape each survivor into one testable sentence

An idea you cannot state crisply cannot be tested. Force each one into: "For [a specific person] who [has this problem], [the thing] does [what], better than [what they do today], because [your advantage]." Writing that sentence is diagnostic — where it goes vague is exactly where the idea is vague. "Everyone" in the first slot is the classic warning sign; an idea for everyone is an idea for no one in particular, which is who buys it.

6. Do not fall in love yet

Affection for an idea grows with time spent on it, not with evidence for it — which is why the longer you polish one in private, the harder it becomes to hear a no. Hold several candidates loosely instead of one tightly, decide what evidence would promote or kill each, and let the next stage — building the smallest testable version — do the choosing.

The whole thing, on one line

Collect problems → spot the workarounds → know your unfair advantages → generate twenty before judging one → shape each into a testable sentence → and let evidence, not affection, pick the winner.