Finding out that somebody is already doing it is the good result.
A competitor has already paid for the expensive discoveries and published them for you. That people have this problem. That they will pay money to solve it. That they can be found and sold to. Those are the three things most ideas die for want of, and a working competitor settles all of them before you have spent anything.
What their existence does not settle is whether they are doing it the way customers want. That is the part left for you, and there is a reliable place to look for it:
Everyone already in the market is competing on the same thing. Their customers are usually complaining about something else. The gap between those two is where your idea goes.
1. Search properly, and hope you find something
Spend an hour genuinely looking. Search the words a customer would use rather than the words you use — people search for the problem, not for the category you have invented. Check app stores, Product Hunt, the relevant subreddit, and what shows up in the ads on those searches, because somebody paying for a keyword is somebody who has already found this market worth money.
If you find competitors, good. Go to step two.
If you genuinely find nothing, slow down. An empty market usually means one of three things, and only one of them is good news:
- Nobody has the problem badly enough to pay. The most common, and the cheapest to discover now.
- Somebody tried and it did not work. Search for the dead ones too — a shut-down product with a farewell blog post is one of the most useful documents you will ever read about your idea.
- You are early. A real position, and an expensive one, because you pay to teach the market something it does not yet know.
2. Write down what the customer is actually trying to get done
Before you look at any competitor closely, write down what the customer wants — not the features they use, the outcome they are after.
Someone using an invoicing tool is not trying to send invoices. They are trying to get paid without spending their Sunday chasing people. Someone using a meeting recorder is not trying to have a transcript. They are trying to stop taking notes while also paying attention.
Do this first and in your own document, because once you start reading competitor websites their language will replace yours. Every market has a house vocabulary, and it is designed to make you think about the product the way the incumbents already do — which is the exact frame you are trying to get outside of.
3. Work out what they are all competing on
Now read the competitors. Not for features — for claims. Put the homepage headline of every one of them in a list.
What you are looking for is the thing they all say. Every market settles on a shared idea of what matters, and once it does, everybody races along that same line: faster, more accurate, more integrations, cheaper. When five companies all lead with the same promise, that promise is the industry's answer to "what is this product for."
Two things are worth noticing. The first is what they compete on. The second, and more useful, is what none of them mentions at all — the decisions so settled that nobody thinks to defend them. Those are not features. They are assumptions, and they are usually invisible from inside the industry.
how much it settles
- People will pay money for this
The single most expensive thing to find out, already answered.
- There is a group you can describe and reach
They found them. You can see how, by looking at where they advertise.
- Roughly what it is worth
Their pricing is a market's answer, arrived at over years.
- How the job is usually done today
Which is what you are competing against, more than the company itself.
- That the space is full
It proves nothing of the kind. This is the one founders read into it.
The bottom bar is the fear that stops people. The top four are what they get in exchange for it.
4. Go and read what their customers complain about
This is the half that most people skip, and it is where the answer is.
Go where customers talk to each other rather than to the company. Two- and three-star reviews on app stores and G2. The subreddit for the job rather than for the product. Posts about switching from one tool to another. Support forums, where the same question gets asked forty times.
Three rules make this much more useful:
- Discount the one-star reviews. They are usually about billing, an outage or one bad support experience. Anger is not the same as insight.
- The best complaints come from people who still use it. Somebody who has stayed for two years and is still annoyed has found something structural, not a reason to leave.
- Look for workarounds, not requests. "I wish it had X" is a feature request. "I export it to a spreadsheet every Friday and re-do it by hand" is somebody paying a weekly tax, which is a much stronger signal.
You are reading for one thing: the complaint that keeps coming up and that none of the competitors has addressed. If you have already worked out where these people gather, you know where to read. If not, that comes first, and it costs nothing but an afternoon.
5. Look for the thing everyone assumed
Now put the two lists next to each other: what the companies compete on, and what the customers complain about. Where they do not overlap is where your idea goes.
The clearest recent example is AI meeting notes. By 2024 the category was crowded — Otter, Fireflies, Zoom's own assistant, and a long tail of others. They competed on the things you would expect: transcription accuracy, summaries, integrations, how many platforms they supported.
Underneath all of it sat a design decision none of them treated as a decision. Every one of these products worked by sending a bot to join your meeting. It appeared in the participant list. Everyone on the call saw it arrive. That was simply how the category worked, and no competitor was competing on it because no competitor considered it a variable.
The customers were not silent about it. What people complained about was not accuracy. It was the awkwardness of a machine turning up uninvited to a conversation with a client, and having to explain it, and not knowing whether anyone had agreed to being recorded.
Granola launched in May 2024 with the bot removed. It takes audio from your own computer instead, so nothing joins the call and nobody else sees anything. The company still leads with it today — its site describes the product as "notes, actions and memory. Without a meeting bot." The absence is the pitch.
Everything since has confirmed the complaint was real and large. In December 2025 a class action was filed against Fireflies in Illinois over collecting voiceprints without written consent, and Otter faces a federal class action in California over recording participants who never agreed. A survey of 500 employed US adults in July 2026 found a third had sat in meetings with an AI notetaker present — and among those, only 34.7% said they were always asked first. Granola went from a $250M valuation in May 2025 to $1.5B in March 2026.
The angle was not a better transcript. It was noticing that the entire industry had agreed on something its customers had not.
Is your angle a gap or a wish?
The distinction that decides whether this is worth spending money on. Answer for the angle you have now, not the one you would have after more research.
Where did your angle come from?
If you described your angle to the companies already doing this, what would they say?
How many people have you heard raise this without being asked?
Answer for the angle you actually have
Not the one you would have after another month of research. The point is to find out now whether there is something here, while changing your mind is still free.
6. Write the angle in one sentence, then go and test it
An angle you cannot say in a sentence is not an angle yet. The shape that works: for [these people], unlike [the existing options], this [does the thing they are complaining about].
Then treat it exactly as you would any other idea — as a starting point rather than a commitment. Being right about the gap does not make you right about the solution, and the market will have opinions about both.
You have an angle when
Then the sequence is the same as for any idea: put the promise on a page, buy a small amount of traffic, and see whether the people complaining about the existing options will act on your version. That last part matters — a complaint is not the same as an intention to switch, and the only way to tell them apart is to ask for something.
When a competitor really is a problem
Competition proves that demand exists. It does not prove you can win, and there are situations where a better angle is not enough.
When the advantage is structural rather than product. Network effects, exclusive supply, a regulatory licence, a distribution deal you cannot replicate. If the reason customers stay is not the product, then improving the product does not move them. The answer is a different market rather than a better feature.
When the complaint is one they have already decided to accept. Sometimes an industry assumption is load-bearing — remove it and the economics stop working. Before you celebrate a gap, spend an hour on why nobody has filled it. If the answer is "because it costs more than customers will pay", you have found a reason, not an opening.
When your angle is really a preference. "Simpler", "cheaper" and "nicer" are the three most common non-angles. If nobody is complaining about the current option being complicated, expensive or ugly, you have found something you would prefer rather than something a market is missing.
Related questions
If you are still choosing between ideas, that comes first: how to decide on a business idea. The people you found complaining are worth checking as a group in their own right — how to tell whether your target market is real. And once the angle is written down, it needs testing rather than building: the whole demand test end to end. And if the worry underneath all this is that somebody will take the idea, that is worth settling early: can you trademark a business idea.
Draper does this part with you — finding who is already serving these customers, what they all claim, what the customers say back, and where the space between them is. Try Draper free →
