How to Come Up With a Business Idea: 7 Ways, a Filter and a Demand Test
To come up with a business idea, start from a problem you or the people near you actually hit. Log the problems you and the people around you run into, add more from the seven sources below, and keep the painful ones that come up often and that people already pay to fix. Then score your best three on five criteria and test demand for the winner before you spend money.
- Log problems for 21 days.
- Mine seven sources.
- Filter: frequent, painful, paid for, fits you.
- Score three ideas 1–3 on five criteria.
- Test demand against a pass line.
The method in five lines:
Pick your starting point:
- No idea at all: start the problem log.
- A problem, no idea: go to ask whether there is an easier way.
- An idea already: jump to the filter.
If you want a starting list before you log anything, find ideas that fit your budget and skills.
Start with a problem you can see
Paul Graham, in his essay “How to Get Startup Ideas,” advises looking for problems, preferably problems you have yourself, instead of trying to think up startup ideas2. A problem gives you something to watch and count. An idea in your head gives you nothing to measure until you build it.
The method on this page narrows in five stages. You collect problems from seven sources into a log and filter them with four questions. The survivors get a score, and the winner gets a test. Each stage drops most of what enters it, so your money goes only to the idea that passed all five.
Plenty of people want to start. LendingTree found that 27% of Americans seriously considered starting a business in the past 12 months, and the figure is 51% among Gen Z3. Wanting to start and having a problem worth solving are different things, and the stages below separate them.
Ready-made lists show what kinds of businesses exist, so browse business ideas if you want that view. Treat any idea from a list as a hypothesis. It still goes through the filter and the score, because a list knows nothing about your buyers or your budget.
7 ways to come up with a business idea
Use one source or several. Each method below has an example and an exercise you can finish in an evening.
- Keep a 21-day problem log
- Ask whether there is an easier way
- Improve one lever of a business you already use
- Follow the money people already spend
- Read the bad reviews of popular products
- Use what you know from the inside
- Move a working model to a new place or customer
Choose sources that suit how you want to work. Your skills, your goals, your location and whether you want to sell online or in person all come back in the “fits you” question of the filter.
Keep a 21-day problem log
Antler’s guide to startup ideas recommends a 21-day notebook where you write down every problem that frustrates you, at work and at home1. Write the problem as it happened, with no solution attached. “I lost a morning waiting for a plumber who never confirmed a time” is a log entry. “An app for plumber scheduling” is an idea, and it comes later.
Copy these five columns into a notebook or a spreadsheet:
| Date | Problem | Who else has it | What it cost you | Times seen |
|---|
Add a tally mark each time the same problem comes back, whether you hit it yourself or hear it from someone else. Our rule of thumb: a problem seen 3 or more times in the 21 days moves to the filter. One-off annoyances stay in the log.
Exercise: start the log today and set a reminder for the evening of day 21.
Ask whether there is an easier way
Clayton Christensen’s jobs-to-be-done theory says people “hire” products to get a job done4. Ask which job someone is trying to finish and where the current way of doing it wastes their time or money. Our guide to jobs to be done covers the theory in full.
Notarize started from that question. Getting a document notarized meant finding a notary and showing up in person, and Pat Kinsel built the company to make that easier5.
Exercise: pick one task you did this week, such as renewing a permit or returning a parcel. Write each step on its own line and mark the slowest one. If the slowest step is one people would pay to skip, you have a candidate. Many service business ideas come from exactly that step.
Improve one lever of a business you already use
You do not need a new category. A business usually competes on one of four levers:
| Lever | Example change |
|---|---|
| Delivery | the dry cleaner picks up at your office |
| Location | the service comes to a place people already are |
| Cost | a smaller package at a lower price |
| Customer experience | booking in one message instead of a phone call |
Keep a running list of businesses you pay and like. For each one, ask what it would look like with your own stamp on one lever.
Exercise: take three businesses you paid this month. Write one lever you would change for each and who would care about the change.
Follow the money people already spend
Money already changing hands is the clearest sign of a real problem. Look for four signals:
- People pay someone to do it.
- They patch it with a workaround, like a shared spreadsheet or a borrowed tool.
- They build a DIY fix.
- They buy something they complain about while paying for it.
Talk to the people you know, from family to coworkers, but ask about what they did. “What did you pay someone to do last month?” gets a fact. “Would you buy this?” gets a polite yes that predicts nothing.
Exercise: ask three people what they paid for last month and hated paying for. Write each answer in your log with the amount they name.
Read the bad reviews of popular products
When a product sells well but rates poorly, people want it and are not getting what they want. Read the one- to three-star reviews of popular apps or local services in a category you know. Copy each complaint that repeats into your log, in the reviewer’s words.
For a wider view of a category, run a SWOT analysis on its leading companies and look at the weaknesses column. Our guide to customer pain points shows how to group and rank the complaints you collect.
Exercise: pick one category and read reviews of its three most popular products until the same complaint shows up three times.
Use what you know from the inside
Years inside a trade or community show you problems outsiders never see, like the form everyone fills in twice or the tool that breaks every season. You also know the words buyers use, which makes the later test cheaper.
One warning. Your own annoyance is a lead to follow up, and it is not proof that a market exists. Count other people who share it before you move on.
Exercise: list three jobs or communities you know from the inside. Next to each, write one complaint you have heard more than once.
Move a working model to a new place or customer
A business proven in one city can often be copied in another, or moved to a new type of customer. A service sold to homeowners can move to small offices. A product built for one hobby can serve a neighboring one. The model is already tested; your question is whether the new buyers have the same problem.
If you want to work from a laptop, our list of online business ideas shows models you can carry to a new audience.
Exercise: name one business that works where you have lived or traveled and is missing where you are now. Write who would buy it here.
Filter each problem with four yes or no questions
Every problem that reached 3 sightings now gets four questions. A problem passes only with four yes answers.
| Question | How you answer it | Pass sign |
|---|---|---|
| Is it frequent? | count tally marks; ask others when it last happened | 3 or more sightings in 21 days, or people name a recent date |
| Is it painful? | ask what it cost the last time | a lost hour, lost money or a missed deadline, named without prompting |
| Is it already paid for? | ask what people do about it now | they pay someone, run a workaround or spend time on a DIY fix |
| Does it fit you? | compare it with your budget, place, skills and time | you can test it with the money and hours you have now |
Pain and payment are separate rows on purpose. People can hate a problem and spend nothing on it, and that problem fails the third question. A problem that passes all four is ready for a value proposition, once you know how you would solve it.
Score your top three ideas on five criteria
Turn your best three problems into three ideas, one solution each. Then score each idea from 1 to 3 on five criteria:
| Criterion | 3 | 2 | 1 |
|---|---|---|---|
| Pain | recurs weekly, costs time or money | recurs, people tolerate it | occasional |
| Free traffic | buyers search or gather where you can reach them free | reachable only with paid ads | you don’t know where they are |
| Money | people pay for a workaround | they spend time or DIY it | no sign of spending |
| Difference | a named rival lacks something buyers complain about | you differ on one lever | same as the closest rival |
| Confidence in the data | you saw it or heard it from buyers | second-hand: reviews, forums | your guess |
These are the five criteria of the matrix we use for business concepts. The idea-stage reading of “money” is ours: before you have prices, it means evidence that people spend.
The rules:
- The maximum is 15.
- A 1 on money or on confidence means you test that idea before you choose it.
- On a tie, the higher confidence score wins, then the higher money score.
- Never raise a confidence score without new evidence.
- Once you have prices and costs, re-score money as LTV/CAC, the lifetime value of a customer divided by the cost to win one. A ratio of 3 or more scores 3, from 1 to under 3 scores 2, and under 1 scores 1. Work out the inputs with our unit economics calculator.
The confidence criterion keeps you honest. An idea that scores 13 on guesses is weaker than one that scores 11 on what buyers told you; the tie and test rules show this.
Example: three ideas from one problem log
This is an example by our method. No real check sits behind it, so every figure in this section is an assumption.
You keep the log for 21 days, the length Antler recommends1. It ends with 40 entries (assumption). Three problems show up 3 or more times (assumption): cyclists at work who never find time for a bike tune-up, parents juggling youth-sports schedules, and a small landlord friend buried in repair requests. All three pass the filter, and you turn them into three ideas.
| Idea (scores are assumptions) | Pain | Traffic | Money | Difference | Confidence | Total |
|---|---|---|---|---|---|---|
| A: mobile bike tune-ups in office parking lots | 2 | 3 | 3 | 2 | 2 | 12 |
| B: schedule coordination for youth-sports parents | 3 | 2 | 1 | 2 | 2 | 10 |
| C: maintenance-request handling for small landlords | 2 | 1 | 2 | 2 | 1 | 8 |
Idea A leads with 12 points (assumption). The cyclists you asked already pay a shop, so money gets a 3 (assumption). They work in buildings you can walk into, so traffic gets a 3 (assumption). Confidence stays at 2 (assumption) because you heard about the problem from coworkers, not yet from paying buyers.
Idea B has the highest pain score, a 3 (assumption), and the lowest money score, a 1 (assumption). Parents complain, but the ones you asked use a free group chat and pay nothing. The rule says test B before choosing it.
Idea C scores 8 (assumption), with a 1 on confidence (assumption), since your only source is one friend. It also goes to a test before any choice.
A has no 1 and the highest total, so you test A first. The test in our example: a one-page booking site plus flyers at 2 office buildings for 14 days, on a $150 budget (all assumptions). The pass line is 10 bookings or more, and the fail line is under 3 bookings (assumption). You write both lines down before the flyers go up.
Test demand before you spend
A demand test asks buyers to take a real step, such as booking, signing up or paying a deposit. Set six things in writing before you start:
- A hypothesis: who will do what, and where.
- One metric that counts the step.
- A budget you can afford to lose.
- A duration with an end date.
- A pass line: the count that sends you forward.
- A fail line: the count that sends you back to the log.
A result between the two lines means you change one thing, such as the price, and run the test again. Changing several things at once leaves you unable to tell which one worked.
Match the test to what you sell:
| If you want to sell | Test it by |
|---|---|
| A local service | flyers plus a booking page with real slots |
| An app or online product | a landing page with a sign-up or pre-order button |
| A course or guide | a presale page with refundable payment |
| A physical product | a pre-order page with photos of a prototype or mockup |
| A service for businesses | a one-page offer sent to businesses that fit, asking for a call |
| A marketplace | matching buyers and sellers by hand for the first deals |
A button for a product that does not exist yet is a fake door test; that guide covers the setup and the ethics. If you talk to people during the test, use our customer interview questions, which ask about past behavior. When you are ready, run a demand test on the idea that scored highest.
Mistakes that sink an idea before it starts
| Mistake | Fix |
|---|---|
| Waiting for an original idea | improve one lever or move a working model |
| Waiting at all | start the 21-day log this week |
| Loving the idea | keep the problem and swap solutions freely |
| Asking opinions | ask what people paid for last month |
| Treating your annoyance as a market | count others who share it |
| Building before a test | set a pass line and a fail line first |
Waiting costs more than most people expect. A 2026 survey published by Founder Reports found that 64% of US adults have at least one business idea they never launched6. Among people who did start, 47% said they should have started years earlier, according to the ZenBusiness State of Entrepreneurship Report 2026 as reported by Forbes7.
Starting with a weak idea and testing it beats holding a good idea you never test. Founder Institute puts it this way: “ideas are useless until they are actually implemented”8. The idea is the first step of a business concept, which builds the customer and the offer around it and puts numbers on them.
Once an idea has passed every stage up to a demand test, check your idea free before you commit money to it.
FAQ
I want to start a business but have no ideas. What should I do?
Start the 21-day problem log1 and add every frustration you or others mention. If you want a list to react to while the log fills up, find ideas that fit your budget and skills.
I have no product and no special skills. Where do I start?
Follow the money people already spend and look for tasks they pay someone to do. Services need no product, and moving a working model to a new place needs no invention. Our side hustle ideas list models you can start with little.
Does a business idea have to be original?
No. Buyers pay for a difference they can name, such as faster delivery, a better location, a lower price or an easier experience. Improving one lever of an existing business or moving a model to new customers both count.
How do I know if my business idea is good?
It passes the four filter questions, scores high on the five criteria with no 1 on money or confidence, and clears the pass line of a demand test. Then check your idea free.
Is coming up with a startup idea different from a small business idea?
The start is the same: a problem you can see, filtered and scored by the same rules. A startup idea also needs a test showing that buyers beyond your first group have the same problem.
Do I need money to come up with or test an idea?
Finding ideas costs nothing but time. You set the test budget; the $150 in our example is an assumption. If friends or family put money in, agree in writing whether you owe it back or they own part of the business; if neither, record it as a gift9. For setup and funding, read how to start a business.
Sources
- antler.co, 10/23/2026
- inc.com, 10/23/2026
- lendingtree.com, 10/23/2026
- online.hbs.edu, 10/23/2026
- online.hbs.edu, 10/23/2026
- founderreports.com, 09/24/2026
- forbes.com, 10/16/2026
- fi.co, 10/23/2026
- stripe.com, 10/23/2026
Check your idea for free
Describe your idea and the check collects demand, competitors, complaints and prices, with sources.
