Where Startup Ideas Come From: Problems, Insights and Trends

Startup ideas don't arrive from nowhere. When you look at how real companies started, the origins fall into a handful of repeating patterns: a founder hits a problem and can't stop thinking about it, someone with deep expertise sees what outsiders miss, or something in the world changes and an old way of doing things stops making sense.

This article maps those sources. It covers the "scratch your own itch" route, domain expertise, technology and regulatory change (the "why now" question), Peter Drucker's seven sources of innovation, and Paul Graham's advice on noticing ideas rather than inventing them. It ends with a way to screen an idea before you spend time validating it.

If you're still deciding whether you're even at this point, the idea stage of a startup explains what founders do there. This piece is about where the idea itself comes from.

Noticing versus thinking up

The most useful distinction in the whole topic comes from Paul Graham of Y Combinator. In his essay "How to Get Startup Ideas," he argues that the way to get ideas is not to try to think of startup ideas. The verb you want, he says, is not "think up" but "notice." Y Combinator calls ideas that grow out of the founders' own experiences "organic" startup ideas.

His reasoning is that invented ideas tend to sound plausible and still be wrong. Graham's own example is a 1995 company that tried to put art galleries online, only to learn that galleries didn't want to be online because it wasn't how the art business worked. He says by far the most common mistake startups make is to solve problems no one has, and that this one cost him six months.

That framing shapes everything below. Most recognised sources of ideas are really different ways of putting yourself where problems become visible.

Source 1: Your own problems

"Scratch your own itch" is the oldest advice in the field, and it has a practical logic. If you have the problem, you know it exists. You also understand how painful it is, how people work around it today, and what a decent solution would need to do.

Graham's version of this is the line live in the future, then build what's missing. The idea is that if you spend time at the edge of a field, using tools and methods most people haven't adopted yet, you'll run into gaps. Those gaps are candidate ideas. He describes the best ones as things the founders want, that they can build themselves, and that few others realize are worth doing.

A few conditions make this source work:

• The problem is frequent, not a one-off annoyance.
• Other people in a similar role have it too. One person's quirk isn't a market.
• You can describe how you cope today, because that's the competition.

The weakness is that your own problem can be unrepresentative. You're one data point. Founders sometimes mistake "I want this" for "a group of people will pay for this." That's why every idea from this source still needs conversations with other people, which is the work covered in customer discovery.

Source 2: Domain expertise and unfair insight

The second source is knowing a field well enough to see what's broken. A nurse, a freight broker, or an accountant notices friction that an outsider would never see, and also knows which friction people would pay to remove.

Y Combinator's Kevin Hale frames this as an "insight," the reason your solution will work and why you'll win. In his Startup School talk he lists five kinds of unfair advantage, the first being the founder: are you one in 10 of all the people in the world who can solve this problem? He adds a sobering note that most applicants he sees don't meet that bar. Being a product manager at a big company, he says, isn't rare, because there are a lot of them there.

So expertise counts as a source when it's specific and uncommon. "I worked in logistics" is weak. "I ran customs clearance for a freight forwarder and know exactly where the paperwork breaks" is stronger, because it points at a concrete, narrow problem.

Source 3: Technology and regulatory change (the "why now")

Some ideas aren't new problems. They're old problems that just became solvable. Something changed in technology, regulation, cost or behavior, and a business that couldn't work before now can.

Graham makes this point when he says the most successful startups generally ride some wave bigger than themselves. He suggests asking what new things become possible as costs fall, and what we're unconsciously ruling out as impossible that soon won't be. He treats this as a fallback to the organic method, a way to simulate noticing when you haven't lived at the frontier.

This is what investors mean by "why now." A good answer names the specific change and explains why it opens a door today and not five years ago. Typical triggers include:

• A technology crossing a cost or quality threshold.
• A new rule that forces companies to do something they used to ignore.
• A shift in how people behave, such as new habits around remote work or online payments.
• A platform opening up that gives small builders access to something only big players had.

The risk with this source is crowding. A visible wave attracts many founders at once, so the idea itself is rarely rare. Your edge has to come from execution, a narrower focus, or some of the other advantages in this article. For more on the timing question, see market timing.

Source 4: Drucker's seven sources of innovation

Peter Drucker approached the question from management rather than startups. In his book Innovation and Entrepreneurship, and in his Harvard Business Review article The Discipline of Innovation, he argued that innovation is a disciplined practice rather than a flash of insight. He says successful innovators don't wait for ideas to strike like a lightning bolt. They go out looking for opportunities in seven key areas.

The first four sit inside a company or industry. The last three come from outside it, in broader social or demographic trends. The HBR piece gives an example for each.

Two things are worth taking from this list. First, most of the sources are about noticing change or friction, which fits Graham's "notice" advice. The unexpected and the incongruous are things you spot by paying attention to customers and operations, not by brainstorming. Second, Drucker is clear that new knowledge is the glamorous source and the hardest. He notes it requires long lead times and the convergence of different kinds of knowledge.

For a founder, the practical use of this list is as a checklist for observation. Instead of staring at a blank page, ask where you've seen a surprise, a mismatch, a clumsy process, a shifting market, a population trend, a change in mood, or a new capability in the field you know.

Key Facts: Sources of Startup Ideas

• Paul Graham says the verb to use with startup ideas is not "think up" but "notice", and calls ideas that grow out of founders' own experiences "organic."
• Graham says by far the most common mistake startups make is to solve problems no one has.
• Drucker's HBR article names seven key areas for innovation opportunities: unexpected occurrences, incongruities, process needs, industry and market changes, demographic changes, changes in perception, and new knowledge.
• Kevin Hale of Y Combinator lists five types of unfair advantage: founder, market, product, acquisition and monopoly.
• Hale's founder test asks whether you're one in 10 of all the people in the world who can solve this problem.

Source 5: Hunting for ideas on purpose

Sometimes there's no obvious itch. Graham addresses this directly and admits that deliberately searching is plan B. His suggestions are about changing your position rather than forcing a brainstorm: work on interesting projects, spend time with people who do, and let gaps accumulate. He also names two filters that hide good ideas, the "unsexy" filter and the "schlep" filter. Most people unconsciously skip ideas that involve tedious, messy work, and he argues that valuable ones are sitting there waiting to be implemented in that territory.

Another deliberate route is to look at what Graham calls small markets that big players ignore. He notes that startups that consume incumbents usually start by serving some small but important market the big players ignore, and offers Steve Wozniak's first computer as an example, which Hewlett-Packard turned down. This ties to the idea of a beachhead market, a narrow starting segment you can win before expanding.

How the sources compare

No source is best. Each has typical strengths and typical traps, and the right question is which one you're drawing from and what extra evidence it needs.

A hypothetical example

Here's a made-up scenario to show how sources combine. Imagine a former hospital billing coordinator who is tired of re-entering the same insurance codes into three systems (own problem and expertise). She learns that a new reporting rule will force small clinics to submit data in a standard format next year (regulatory change). The itch makes the problem real to her, the expertise narrows it to a specific buyer, and the rule supplies the "why now."

Notice what's still missing: she hasn't shown that clinics would pay, or that her way of solving it beats a spreadsheet. That's the point of the next section.

Evaluating an idea before you validate it

Validation means going out and testing the idea with real people. Before that, a quick desk check can rule out ideas that clearly won't hold up and save weeks. Kevin Hale's framework is a useful structure. He describes a startup idea as a hypothesis about why a company could grow quickly, built from three parts: a problem, a solution, and an insight.

On the problem, he says good problems are popular, growing, urgent, expensive to solve, mandatory and frequent, and a problem needs at least one of those traits, ideally several. On the solution, his advice is blunt: don't start here. YC has an acronym for the failure, SISP, a solution in search of a problem, where an engineer excited by a new technology goes looking for something to apply it to. On the insight, he asks for an unfair advantage that explains why you'll grow faster than everyone else.

A short screening checklist built from the ideas in this article:

1. Source. Which of the sources above did this come from? If it's pure brainstorming, treat it with extra suspicion.
2. Problem. Can you name who has it, how often, and what they do today?
3. Why now. What changed that makes this possible or urgent?
4. Insight. Why you and why this team? Which of Hale's five advantages could you plausibly claim?
5. Size. Is there a narrow first segment and a believable path to something larger? TAM, SAM and SOM give the vocabulary.
6. Biggest unknown. What single belief, if wrong, kills the idea? That belief is what you test first, as in a riskiest assumption test.

This is a screen, not proof. It tells you whether an idea deserves testing. It can't tell you that people will buy.

What to do after you have a candidate

Once an idea passes the desk check, the work changes from thinking to evidence. Two steps usually follow:

• Check problem-solution fit. Find out whether the problem is real for a defined group and whether your approach addresses it. The article on problem-solution fit covers this.
• Talk to customers. Customer discovery is the practice of interviewing potential users about their past behavior, not their opinions of your idea.

Frameworks like Jobs to Be Done help frame those conversations around the progress customers want to make. And if you're early in the company's life, the stages of a startup show where idea sourcing sits in the larger sequence.

Common mistakes when sourcing ideas

• Brainstorming in isolation. A list of clever ideas written at a desk is the thing Graham warns about.
• Treating one anecdote as a pattern. A single surprising event is a clue, not a market.
• Chasing the loudest wave. If everyone sees the same trend, the trend alone won't differentiate you.
• Skipping the unglamorous. The schlep and unsexy filters quietly remove ideas that might be good because they're tedious.
• Falling for the solution. Starting from a technology you like is the SISP trap.
• Confusing expertise with advantage. Working in an industry isn't the same as having an insight others lack.

Related reading

• Idea stage of a startup
• Problem-solution fit
• Customer discovery
• Lean startup method
• Jobs to Be Done
• Market timing