To build a startup from scratch, start with a real problem that many people share. Talk to those people before you build anything, then make the smallest product that tests one risky assumption.
Get it in front of a handful of users, watch what they do, and only then think about money, hiring or scale. This guide walks through each stage in order, with the checks that stop you wasting months.
Let's jump in.
TL;DR
- Start from a widespread problem or market gap, not from a product idea or a personal annoyance that may already be solved.
- Validate with direct conversations, not broad surveys. Look for people who already feel the pain and are working around it.
- Build an MVP that is the cheapest credible experiment: one core workflow, one written hypothesis, one decision rule.
- Find your first users in the places they already gather, then refine the product from what they do, not what they say.
- Match funding to the business you are building: bootstrapping keeps control, venture capital trades equity for speed.
Step 1: Choose a problem worth a few years of your life
Most failed first startups begin with a solution looking for a problem. Reverse that. Guidance from JPMorgan says a startup should begin by identifying a genuine, widespread problem or market gap, rather than focusing solely on a personal inconvenience that might already be solved.
That last part matters. Scratching your own itch is a fine starting point, but check that the itch is common and that nobody has fixed it well already.
A few questions help you screen ideas quickly:
- Who exactly has this problem? If your answer is “everyone”, you have not decided yet. Name a job title, a type of shop, a hobby.
- What do they do today? A spreadsheet, a clumsy workaround or a paid agency all count as evidence. If they do nothing, the problem may not hurt enough.
- Do you have an unfair advantage? Experience in the industry, an audience, a technical skill or a network all make the first year cheaper.
- Can you reach them? A problem you cannot reach customers for is a hobby, not a business.
Write your answers in one page. If you cannot fill it, you have homework, not a business yet.
Step 2: Validate before you build
Validation means finding out whether people will actually pay, and doing it before you spend months coding. Stripe’s startup guide frames market research as verifying demand through targeted conversations rather than broad surveys, and as checking online communities such as Reddit, or search volume tools, for problems that keep coming up.
Here is how that works in practice.
Talk to ten people, not a thousand
A survey tells you what people say they might do. A conversation tells you what they did last week. Aim for ten to fifteen short calls with people who match your target. Ask about the last time the problem happened, what they tried, what it cost them and what they would do if it vanished tomorrow. Avoid pitching. The moment you describe your idea, people start being polite.
The JPMorgan guide, linked above, also points to talking to potential customers to gauge buyer intent. Buyer intent is the useful signal: someone asking when it will be ready, offering to pay early or introducing you to a colleague is worth more than ten compliments.
Read where your customers already complain
Search niche forums, subreddits, Slack groups and review pages for the words people use about the problem. Note the repeated phrases. They become your landing page copy and your search keywords later. Search volume tools show whether people look for a fix at all, which is a cheap sanity check on size.
Check the competition honestly
Existing competitors are usually good news: they prove people pay. Map the alternatives by type: direct rivals, adjacent tools, and the manual workaround. Then write one sentence on why a specific group would switch to you. A business plan that outlines your objectives, target market, products and services and competition, as JPMorgan describes it, forces you to put that sentence on paper.
Step 3: Build a minimum viable product that tests something
An MVP is not a cheap, worse version of your dream product. It is an experiment. One product guide describes it as the cheapest credible experiment to test a specific hypothesis, centred on a core workflow and leaving out secondary features such as advanced billing or complex reporting.
Write the hypothesis first
Before development, write down the user outcome you expect and a decision rule that tells you whether the test worked, as the same guide recommends. For example:
If 20 freelance designers try our invoice tool, at least 8 will send a real invoice within a week. If fewer than 5 do, we change the onboarding or the audience before we add features.
The numbers are yours to choose. The point is to agree them before you see the results, so you cannot talk yourself into a pass.
Scope to the primary workflow
An MVP should include only the core features needed to complete the primary user workflow and test a specific, risky hypothesis, according to SuperMe’s MVP scoping guide. List every feature you imagined, then cross out everything that is not needed to get one user from problem to result.
You may not need code at all. A landing page with a waiting list, a clickable prototype, a manual service you perform behind the scenes, or a no-code build can all test demand. If you want to see what is available at no cost, our list of free tools for bootstrapped founders is a sensible place to start, and prototyping with AI 3D generation shows how physical-product founders are cutting early costs.
Test behaviour, not opinions
Watch people use the product. SuperMe’s guidance is to observe actual behaviour and focus on whether users complete the core task, rather than just asking what they think. Sit with a user, share your screen, say nothing and note where they hesitate. Five sessions usually reveal the same two or three problems.
Step 4: Get your first customers
Your first users will not come from an advert.
Stripe suggests targeting early adopters and using social media, niche forums, networking and a referral programme, then tracking engagement and refining the MVP from early feedback.
A simple sequence for the first ninety days:
- Go back to your interviewees. People who described the problem are your warmest leads. Offer them early access.
- Post where the problem is discussed. Be useful first. Answer questions, share what you learned, and mention the product only where it genuinely fits.
- Do things that do not scale. Onboard each user personally. You learn faster and they feel looked after.
- Add a referral nudge. Ask each happy user who else has the same problem.
- Track a few numbers. Sign-ups, completion of the core task and returning users tell you more than page views.
If you want low-budget ideas for getting attention, our piece on guerrilla marketing tactics for startups has examples, and building brand authority online covers the longer game.
Decide what to prioritise
Feedback will always outrun your capacity. Sort every request by two questions: how many users asked, and does it help them finish the core task? The 80/20 rule, or Pareto principle, is a useful mental model here: a small share of your effort tends to drive most of your results.
In a young startup that usually means a few features, one or two acquisition channels and a small group of customers. Find them, protect your time for them, and say no to the rest for now.
Step 5: Sort out the basics of running a company
Before money changes hands, get the foundations right. Decide your legal structure, put agreements in writing, and keep personal and business finances apart. Our guide to the key documents in a startup’s legal foundation lists what to have in place. If you hire, even a first contractor, think about workplace safety from day one too.
Rules differ by country, so confirm requirements with your national company registry or a qualified accountant or solicitor before you register or sign anything.
A one-person setup can run on very few tools. We tested options by job in our guide to the best tools to run a one-person business, and our piece on boosting productivity with smart tools covers the day-to-day.
Step 6: Pick a funding route that fits the business
Funding is a consequence of your plan, not the first step. The US Small Business Administration sets out the trade-off plainly.
Self-funding keeps you in complete control but puts all the risk on you. Venture capital typically targets high-growth companies, swaps equity for capital and may mean giving up some control. The SBA is a US body, so check which loans, grants and schemes exist where you are based.
| Route | What you give up | Often suits |
|---|---|---|
| Bootstrapping | Speed, since you grow from revenue | Service, SaaS and content businesses with fast paying customers |
| Grants and loans | Time spent on applications, plus repayment for loans | Founders with a clear plan and local schemes available |
| Angel investment | Some equity and some control | Early traction, a founder who wants an experienced backer |
| Venture capital | Equity and some control | Businesses built for very fast growth in large markets |
This table is a rule of thumb for orientation, not advice for your particular case.
Many indie hackers start by bootstrapping because it forces discipline. Our guide on what bootstrapping means in business goes through how it works, and our look at what founders actually pay themselves helps you plan your own runway.
Step 7: Find your people
Building alone is slow and lonely. Look for peers a step or two ahead of you: founders who have shipped, sold or failed and will tell you how. Good places include niche online forums, local founder meetups, small mastermind groups and the newsletter communities for your sector.
Offer help before you ask for it, and be specific when you do ask. “How did you get your first ten customers?” gets a better answer than “any advice?”
Reading helps too. Our list of books bootstrappers actually read is a good shelf to start from.
Our perspective
Most startup advice is written for people who already have a team, a runway and a polished idea. The founders we hear from rarely do. They have a day job, a few hundred pounds, and a hunch.
We built We Are Founders for exactly that person. We hand-curate tools and tactics, and we favour advice from peers who have real startup experience, including exits, over polished theory. Our aim is to cut the noise: fewer articles, more of what you can use this week.
Our view is that the first year is about learning faster than your money runs out. That means talking to customers early, building small, and borrowing from people who have already made the mistakes. A good directory of tools, a few honest founder stories and a community of peers shorten that learning curve, and that is what we work to provide.
—Jessica
Common mistakes when starting from scratch
- Building for months before talking to anyone. The fix is the interview step above, done before a line of code.
- Adding features instead of finding users. More features rarely rescue a product nobody has tried.
- Skipping the hypothesis. Without a decision rule, every result looks like encouragement.
- Raising money too early. Outside funding adds pressure. Raise when you know what the money will buy.
- Going it alone. Peers shorten the path and keep you going when progress is slow.
FAQ
How do I start a business from scratch with no money?
Pick a problem you can test without spending much: talk to ten potential customers, then sell a manual version of the service or a pre-order before you build. Free and low-cost tools cover a landing page, invoicing and email. Revenue from your first few customers can then fund the next step, which is the heart of bootstrapping.
Is it true that 90% of startups fail?
The figure is repeated so often that it is worth treating with caution, because definitions of “startup” and “fail” vary widely between studies. A more useful question is why individual startups stall: usually no real demand, running out of cash before learning enough, or a founder stopping. You can act on all three by validating early, keeping costs low and building a support network.
What is the 80/20 rule for startups?
It is the idea that roughly 80% of results come from 20% of causes. For a founder, that means a small number of customers, features and channels usually drive most of your growth. Work out which ones they are, and spend your limited time there.
How long does it take to build an MVP?
That depends on what you are testing, so scope to the hypothesis, not to a deadline. A landing page or manual prototype can take days, a simple software workflow longer. If your plan runs to many months before any user sees it, you have probably scoped too much.
Do I need a business plan?
You need a clear one-page plan covering your target market, product, competition and objectives. JPMorgan’s guide includes a plan along those lines as part of starting up. You can expand it later if a lender or investor asks for detail.
Sources
- 10-Step Guide to Starting Your Startup Business, JPMorgan
- How to start a startup, Stripe
- How to Scope an MVP That Tests Your Idea, SuperMe
- How to Build an MVP That Tests the Right Assumptions, Hapy
- Fund your business, US Small Business Administration
