How to Find Problems People Will Actually Pay You to Solve

How to Find Problems People Will Actually Pay You to Solve

Short answer: look for problems that are frequent, expensive, already recognised by the person suffering them, and already attracting money — even badly spent money. The strongest signal isn’t someone describing a problem. It’s someone describing a clumsy, expensive workaround they’ve built themselves. A workaround is a receipt. It proves the problem hurts enough to spend on.

Most idea generation fails because it starts from solutions. You have a capability, you go looking for somewhere to apply it, and you end up with a business that is technically competent and commercially invisible.

This guide runs in the opposite direction. Find the paid problem first, then decide whether you’re the right person to solve it.

What separates a paid problem from an annoyance

Four properties, and you need at least three of them.

Frequency. How often does the problem occur? Weekly problems build businesses. Once-a-decade problems build very hard businesses, because you have to find each customer fresh at exactly the right moment, and that makes acquisition brutally expensive.

Severity. What does it cost when it happens — in money, time, risk or embarrassment? Put a number on it. “It takes our office manager six hours a month” is severity. “It’s a bit annoying” is not.

Existing budget. Is money already flowing at this problem? Towards a competitor, a member of staff, a piece of software, a workaround, a stopgap. 

Money already moving is the single best predictor that money can move to you. Creating a budget line where none existed is a far harder sale than capturing one that already exists.

Awareness. Does the sufferer recognise it as a problem? Unrecognised problems are the graveyard of clever founders. If you first have to convince someone they have a problem, then convince them you can solve it, then convince them to pay, you’re running three sales processes to close one deal.

The quick filter

Score each property from 1 to 5, then multiply your two weakest scores together. If that product comes to less than 6, walk away — however well the other two scored. A severe problem nobody recognises, or a frequent problem nobody funds, will not carry a business.

Five research methods, and what each is honestly good for

Every method has its own blind spot. Use at least three, because their biases don’t overlap.

1. Review mining — cheapest way to find unmet needs

Go where people rate existing solutions: marketplace reviews, app store reviews, Google reviews for local competitors, industry software review sites.

Read the three-star reviews specifically. Five-star reviews tell you almost nothing. One-star reviews are usually about a single bad experience. Three-star reviews are written by people who genuinely wanted the thing to work and are explaining, in detail, exactly where it didn’t.

Extract four things: the precise phrases people use, what they expected, what they do instead now, and whether they mention price.

Blind spot: reviewers are unrepresentative — they skew towards the most annoyed and the most delighted. Never size a market from reviews.

2. Search and community observation — best for language and volume

People type things into search engines and AI assistants that they would never say aloud to a stranger. Look at question-shaped queries, forum threads, subreddit posts, professional group discussions and related-question expansions.

Two patterns are especially productive:

• “how do I” plus a task — reveals processes nobody has solved properly
• “alternative to” plus a known provider — reveals dissatisfaction with existing spend, which is the best kind of demand, because the budget already exists

Blind spot: online complaint volume massively over-represents categories with young, technical, chronically online users. Trades, care, logistics and manufacturing generate enormous real demand and very little forum chatter.

3. Customer interviews — best for severity and workarounds

The highest-value method, and the easiest to do badly. Ten proper conversations will teach you more than a hundred survey responses.

Questions that work. All of them are about the past, and all of them are specific.

• “Walk me through the last time this happened. What did you actually do?”
• “What did that cost you — in time, money or hassle?”
• “What have you already tried? Why did you stop?”
• “Who else was involved in dealing with it?”
• “What are you currently spending on this, if anything?”
• “If this were solved tomorrow, what would change about your week?”

Questions that will lie to you.

•  “Would you use a product that did X?” Everyone says yes. Saying yes costs them nothing.
•  “How much would you pay for that?” Hypothetical prices are fiction, and usually two to three times the real figure.
•  “Do you think this is a good idea?” You’ve just asked for a compliment, and you’ll get one.

The rule underneath all of it: ask about behaviour that has already happened, never about intentions that haven’t.

Blind spot: people are polite. If everyone you interview is encouraging, you’re either interviewing friends or asking leading questions. Real interviews produce a few flat, disappointing answers. Those are the valuable ones.

4. Frontline observation — best for problems nobody articulates

Spend time where the work actually happens. Shadow someone for a day. Work a shift. Sit in a waiting room. Watch what people do rather than what they say they do.

This is the only method that surfaces problems so normalised that nobody thinks to mention them: the spreadsheet everyone hates, the double entry between two systems that don’t talk, the twenty-minute daily phone call that exists purely to bridge a gap in software.

Blind spot: slow, and a sample size of one. Confirm anything you observe through interviews before you act on it.

5. Payment evidence — best for confirming, worst for discovering

Look for money already changing hands: competitors with paying customers, procurement listings, published price lists, agency retainers — and job adverts.

Job adverts are dramatically underused. A company hiring for a role is a company paying to solve a problem, and the job description is a free, detailed problem specification written by the buyer. If forty local firms are advertising for someone to do X, then X is a funded, recognised, frequent problem.

Blind spot: this only confirms problems someone is already solving. Excellent for validating, useless for finding gaps.

The workaround test

The single strongest indicator of a paid problem is an existing workaround.

Workarounds look like this:

•  A spreadsheet with fourteen tabs that one person maintains and everyone depends on
• A part-time member of staff whose job is essentially manual reconciliation
• Two tools bought separately and glued together by a human being
• A WhatsApp group doing the job of a scheduling system
• Someone driving forty minutes each way because the nearest provider isn’t good enough
•  A recurring subscription nobody likes but nobody has cancelled

Every one of those is a person who has already paid — in money, time or inconvenience — to make the problem smaller. Your offer isn’t asking them to start spending. It’s asking them to redirect spending they’ve already accepted, which is a fundamentally easier conversation.

So in interviews, hunt explicitly for workarounds. When someone says “oh, we just do it manually,” stop and go deep. That sentence is worth more than an hour of hypotheticals.

Worked example: from vague to fundable

Starting point: “Small businesses struggle with admin.” Unusable. No frequency, no severity, no named buyer.

After review mining: three-star reviews of two popular accounting tools repeatedly mention the difficulty of reconciling job costs against invoices for trades running multiple sites.

After eight interviews with small building firms: six of the eight track job profitability in a spreadsheet the owner’s partner maintains at weekends. Two have stopped tracking it altogether and admit they don’t know which jobs make money. 

Time spent averages five to seven hours a month. Two firms describe losing money on a job and only discovering it months afterwards.

After scanning job adverts: nine local firms in the last quarter advertised part-time bookkeeping roles that specifically mention job costing or CIS.

Result: per-job profitability tracking and reconciliation for building firms running three to fifteen concurrent jobs.

•  Frequency: 5 — weekly and unavoidable
• Severity: 4 — six hours a month plus unquantified losses on mispriced jobs
• Existing budget: 5 — already paying part-time staff or an accountant
• Awareness: 4 — they describe it themselves, unprompted
• Weakest two: 4 × 4 = 16. Comfortably through the filter.

Notice what happened there. The problem got narrower at every stage, and more fundable at every stage. Specificity isn’t a limit on your market — it’s what makes the market reachable in the first place.

Five failure patterns worth naming

The vitamin problem. Nice to have, never urgent, permanently deprioritised. The test: has anyone ever dealt with this at eleven at night because it couldn’t wait until morning? If not, it’s a vitamin.

The someone-else’s-budget problem. The person who feels the pain has no authority to spend. Very common in larger organisations and in household purchases. You have to sell to the payer, so make sure the payer feels something.

The free-alternative problem. A good-enough free solution exists and everybody knows about it. You’re not competing with paid competitors, you’re competing with zero, and zero is a difficult price to beat.

The seasonal mirage. The problem is severe for six weeks a year. It’s real, but the business needs eleven months of something else to survive.

The problem you want to exist. The one you’ve decided is important because your skills would solve it beautifully. Guard against this by writing down, in advance, what evidence would make you abandon the idea. If you can’t name any such evidence, you aren’t researching — you’re campaigning.

Your next two weeks

Days 1 to 3. Pick a domain you have real access to. Mine reviews and job adverts. Write down twenty problem statements in the sufferers’ own words, not yours.

Days 4 to 10. Run eight to ten interviews. Ask only about the past. Log every workaround you hear, however small.

Days 11 to 12. Score your top five problems on frequency, severity, budget and awareness. Apply the weakest-two filter honestly.

Days 13 to 14. For whatever survives, write a one-sentence problem statement naming the specific customer and the specific cost. If you can’t name both, you haven’t finished.

At the end of that you have a problem rather than an idea. 

The next question is whether enough people have it — a different question, answered with different tools, in How to Know If There Is Demand for Your Business Idea.

Frequently asked questions

How many interviews do I actually need?

Eight to twelve in a single customer segment is usually enough to see the pattern. The signal to stop isn’t a number — it’s when you stop hearing new things. If your ninth conversation surfaces a workaround you’ve never heard described before, keep going. If conversations seven through ten simply confirm what you already knew, you have your answer and more interviews will only make you more confident rather than more correct.

What if I can’t get anyone to talk to me?

That’s important information rather than a logistical problem. Difficulty reaching a segment for a free fifteen-minute conversation predicts difficulty selling to them later. Before concluding the segment is unreachable, try three things: approach through a shared connection rather than cold, ask for advice rather than an interview, and go where they already gather — trade events, supplier days, professional groups. If all three fail, the access problem is real and it will still be there when you have something to sell.

Do I need to pay people for interviews?

Usually not for consumers or small business owners, who tend to respond well to genuine curiosity about their work. For senior professionals whose time is billable, a modest incentive or a donation to a charity of their choice is reasonable and improves both response rate and candour. What you should never do is pay in a way that makes people feel obliged to be encouraging.

How do I tell a real problem from one I’ve talked myself into?

Write down your kill criteria before you start — the specific findings that would make you drop the idea. Then check whether you’ve actually looked for them. A second test: can you state the problem in the customer’s words, using phrases you heard rather than phrases you invented? If every description of the problem is yours, you’re looking at a hypothesis, not a finding.

What if the problem is real but nobody is paying to fix it?

Then something is blocking the spend, and you need to know what. It’s usually one of four things: the sufferer isn’t the payer, a free alternative is good enough, the cost of the problem is invisible to whoever holds the budget, or previous attempts burned them. The first and third are workable — you can make the cost visible, or sell to the payer instead. The second and fourth are much harder, and generally mean choosing a different problem.

Can I do this research if I don’t work in the industry yet?

Yes, though you’ll lean harder on some methods than others. Review mining, search observation and job-advert scanning need no insider access at all. Interviews are still possible if you approach people as an outsider genuinely trying to understand their work, which many people find easier to respond to than a pitch. What you’ll lack is frontline observation, and that’s the method most likely to surface the problems nobody mentions — so try to arrange even a single day of shadowing before you commit.

Should I tell people about my idea during interviews?

Not until the end, and not as a question. Describing your solution early contaminates everything that follows, because people start being helpful about your idea instead of honest about their situation. Spend the conversation on their past behaviour, and if you want a reaction to the concept, save it for the final two minutes — after you already have what you came for.

Is a problem that only affects a small number of people worth pursuing?

It can be, if severity and budget are high enough to compensate for low frequency. Two hundred firms with a £5,000-a-year problem is a viable business; two hundred thousand people with a £3 annoyance generally isn’t, because reaching them costs more than they’re worth. 

The number that matters isn’t how many people have the problem, but how many you can reach affordably multiplied by what solving it is worth to each of them.