How to find startup ideas: six reliable sources

Stop waiting for inspiration and start looking in fixed places

Brainstorming from a blank page tends to produce the same ideas everyone else produces: a better to-do app, a marketplace for something that already has three, a social network for a hobby. Those ideas are not wrong so much as unanchored. Nothing about them comes from a specific person with a specific problem, so there is nothing to test.

The more reliable approach is to treat idea generation as a habit with fixed sources, the way a journalist works a beat. You look at the same handful of places regularly, you write down anything that snags, and over a few months the log starts to show you which problems keep coming back. Repetition is the signal. A problem you notice once is an anecdote; a problem you notice in four different places is a lead.

The six sources worth working

• Your own work — The tasks you do every week that feel slower than they should, the spreadsheet everyone on your team is afraid to touch, the report you rebuild by hand every month. You already know the vocabulary, the buyer and the workaround, which is an advantage most founders have to spend months acquiring.
• Problems you already pay for — Look at your bank statement and your employer's expense reports. Every recurring payment to a freelancer, agency or clunky tool is evidence that a problem is painful enough to spend money on. Ask whether that service could be cheaper, faster, or turned from a service into software.
• Unbundling — Large products often contain one feature that a specific group uses heavily and finds mediocre. A focused tool that does that one job far better, for that one group, can win without competing on everything else.
• Boring industries — Trades, logistics, property management, local services, compliance-heavy niches. These rarely appear in technology news, which is precisely why their software is often old, generic or missing. Lack of glamour keeps competitors away.
• Regulation and rule changes — A new reporting obligation, a licensing change, or a deadline for a standard creates work that did not exist before, for people who have no tool for it yet. The useful version has a date attached.
• Tool gaps — The handoff between two systems where a person currently copies data by hand. The export that nobody can read. The integration a vendor has promised for years. Gaps between tools are where workarounds live, and workarounds are unpaid demand.

What each source tends to produce

The sources are not interchangeable. Each one has a typical strength and a typical trap, and knowing which you are drawing from helps you judge the ideas it gives you.

How to keep an idea log that actually helps

An idea log is just a single place where every candidate goes, written in a consistent format. A spreadsheet works; so does a plain notes file. What matters is that each entry captures the problem, not the solution, because solutions are cheap to invent and problems are expensive to find.

Use the same fields for every entry so you can compare them later. A useful minimum:

• Date and source — Where you noticed it: your job, a forum thread, a conversation, a news item about a rule change.
• Who has the problem — A role, not a demographic. 'Site manager on small construction projects' rather than 'builders'.
• The problem in one sentence — What goes wrong, in their words if you have them.
• Current workaround — What they do today instead. If you do not know, write 'unknown', which is itself a task.
• Evidence of spend — Any sign of money or significant time going into the workaround.
• Times seen — A simple tally you increase each time the same problem reappears from a different source.

Reviewing the log without falling in love

Set a recurring weekly slot of half an hour to read the log. Merge duplicates, increase tallies, and add any evidence you have picked up since. Resist the urge to start building whatever you wrote down most recently; recency is not quality.

Every month or so, pick the three entries with the highest tallies and the clearest evidence of spend, and give each one a small amount of real investigation: find five people who have the problem and ask how they handle it today. Most entries will fall apart at this step, which is the point. The log exists to make it cheap to discard ideas, so that the one you eventually commit to has survived several rounds of contact with reality.

Say, as an illustration, that you log forty entries over three months. If two or three of those keep reappearing from independent sources and show clear spend, that is a healthy outcome. You do not need a long list of good ideas; you need one that keeps proving itself.

Common ways people sabotage the search

• Starting from a technology — 'What could I build with this new model?' produces solutions looking for problems. Start from the problem and let the technology be whatever fits.
• Only looking at consumer life — Your personal annoyances are valid sources, but businesses pay more readily and more predictably than individuals do.
• Dismissing ideas because competitors exist — Competitors are evidence of demand. The question is whether a specific group is badly served, not whether the category is empty.
• Writing solutions instead of problems — 'An app for scheduling dog walkers' hides the problem. 'Dog walkers lose an hour a day rearranging bookings by text message' is something you can test.
• Never closing the loop — An idea log that is only ever added to becomes a graveyard. Review, test, discard, repeat.

Questions