How to Start a Startup: A Step-by-Step Guide

To start a startup, pick a specific problem that a reachable group of people has, confirm through conversations and small tests that they want it solved, build the smallest product that solves it, and get it in front of paying or committed users as fast as possible. Company formation, fundraising and hiring come after that, in the order your evidence justifies. The rest of this guide walks through each step with checklists and the decisions that trip up first-time founders.

What a startup is (and what it is not)

A startup is a company designed to find a repeatable, scalable business model under uncertainty. That last part is what separates it from a small business: a bakery has a known model, while a startup still has to discover who its customer is, what they will pay, and how to reach them efficiently.

This matters because it changes what "progress" means. In the early months, progress is not revenue or headcount. It is reduced uncertainty: each week you should know more about the problem, the customer and the channel than the week before.

	Small business	Startup
Business model	Known, proven elsewhere	Unknown, has to be discovered
Main early risk	Execution	Demand and model
Typical funding	Savings, bank loan	Savings, revenue, angels, venture capital
Growth expectation	Stable, local	Fast, often across markets
Early metric of progress	Profit	Learning, then traction

Neither is better. But if you call your project a startup, design it to answer the unknowns quickly.

Step 1: Choose a problem, not an idea

Ideas are cheap and usually wrong in the details. Problems are more durable. Start by writing down problems you have seen first-hand, ideally in an industry you know or a role you have worked in.

A good starting problem tends to have these traits:

• Frequent or expensive. People hit it often, or it costs them real money or time when they do.
• Specific owner. You can name a job title or type of person who suffers from it.
• Current workaround. They already solve it badly, with spreadsheets, manual work or a tool they dislike. A workaround proves the problem is real.
• Reachable. You know where these people gather, online or offline, and can get 20 conversations within a couple of weeks.
• Founder fit. You understand the domain well enough to ask sharp questions and spot nonsense.

Write a one-sentence problem statement: "[Who] struggles with [what] when [situation], and today they [workaround]." If you cannot fill it in, you do not have a problem yet.

Step 2: Validate demand before you build

Validation means collecting evidence that people want the problem solved enough to change their behaviour or pay. Compliments do not count. Commitments do: time, money, data, introductions.

A practical sequence:

1. Problem interviews. Talk to 15 to 30 people from your target group about how they handle the problem today. Ask about past behaviour, not hypothetical future purchases.
2. Solution test. Show a mock-up, a landing page or a manual service and ask for a concrete next step, such as a pre-order, a pilot, or a signed letter of intent.
3. Price test. Put a real price in front of people. Their reaction tells you more than any survey.

The full method, including interview questions and pass/fail criteria, is in our guide on how to validate a startup idea.

Step 3: Decide on co-founders and ownership early

Building alone is possible, but the workload of an early startup is broad: product, sales, finance, hiring, operations. Many founders look for a co-founder with complementary skills.

If you start with co-founders, settle three things in writing before you write much code:

• Roles and decision rights. Who has the final say on product, on hiring, on money?
• Equity split. How ownership is divided, and why. Our article on co-founder equity split covers common approaches.
• Vesting. Equity that is earned over time, so a co-founder who leaves after three months does not keep a large stake. See founder vesting schedules.

These conversations feel awkward when everyone is excited. They are much harder once there is something valuable to fight over.

Step 4: Build a minimum viable product

An MVP is the smallest thing that lets a real customer get value and lets you learn whether your core assumption holds. It is not a worse version of the full product. It is a deliberately narrow product that tests one thing.

MVP type	What it is	Good for
Concierge	You deliver the service manually	Testing whether the outcome is valuable
Wizard of Oz	Product looks automated, you do the work behind it	Testing the user experience before automation
Single-feature product	One core workflow, nothing else	Software where the core action is clear
Pre-sale / landing page	Describe the product, take orders or deposits	Testing willingness to pay
No-code prototype	Built with existing tools	Fast iteration without engineering cost

Set a deadline, often a few weeks rather than months, and a single success criterion before you start building. Our step-by-step guide on how to build an MVP covers scoping, tools and what to measure.

Step 5: Set up the company properly

Once you have a co-founder, a customer contract, or an investor interested, it is usually time to formalise. The exact steps depend on your country, but the checklist is similar almost everywhere:

• Choose a legal form that suits your plans (for example, a limited company if you expect outside investment)
• Register the company and get any tax identification numbers you need
• Sign a founders' or shareholders' agreement covering equity, vesting, roles and what happens if someone leaves
• Assign intellectual property (code, designs, domain names) from founders to the company
• Open a separate business bank account
• Set up simple bookkeeping from the first transaction
• Check whether you need permits, licences or data protection measures for your industry

Company law, tax and employment rules differ significantly between countries. Treat this list as a starting point and have a lawyer and tax adviser review your setup, especially the shareholders' agreement and IP assignment.

Step 6: Know your runway from the first month

Runway is how many months you can operate before the money runs out. It is the single most important number for an early-stage founder, because it determines how many experiments you can run.

Example (hypothetical): two founders have €60,000 in savings set aside for the company. Their monthly costs are €1,500 for tools and hosting, €500 for marketing tests and €3,000 in modest founder salaries, so €5,000 in total. With no revenue yet, net burn is €5,000 per month and runway is €60,000 / €5,000 = 12 months.

If they land a first customer paying €1,000 per month, net burn drops to €4,000 and runway extends to 15 months. That shows why early revenue is so valuable even when it is small.

You can model these scenarios, including revenue and cost growth, with the startup runway calculator. Our guide on how to calculate burn rate explains gross versus net burn in more detail, and startup financial metrics covers the other numbers worth tracking from the start.

Step 7: Get your first customers by hand

Early customers rarely come from ads or SEO. They come from direct outreach: emails, calls, introductions, events, communities. This is slow and does not scale, and that is exactly the point. Doing it manually teaches you the language customers use, the objections they raise, and which segment converts best.

A simple weekly routine:

1. List 30 to 50 people who match your ideal customer profile.
2. Send short, personal messages that describe the problem, not your product.
3. Book calls with those who reply and run a demo or offer a pilot.
4. Track every conversation in a simple spreadsheet: who, date, stage, objection, next step.
5. Review the sheet weekly and adjust your message and target segment.

Once a channel works repeatedly, you can start thinking about a broader go-to-market strategy and pricing. If you sell software, read how to price a SaaS product before you set your first price list.

Step 8: Decide whether to raise money

Not every startup should raise venture capital. Outside money makes sense when growth requires spending ahead of revenue, for example on engineering, inventory or a sales team, and when the market is large enough that investors can expect an outsized return.

Ask yourself:

• Can we reach the next milestone with revenue and savings alone?
• Is speed critical because competitors or market windows exist?
• Are we comfortable giving up ownership and some control in exchange?
• Do we have evidence (users, revenue, pilots) that makes the story credible?

If the answers point to raising, read our startup fundraising guide for stages, instruments and process. Before you agree to any terms, run the numbers on how much of the company you would give up with the funding dilution calculator.

Step 9: Make your first hire deliberately

The first people you hire shape the culture and capacity of the company more than anyone after them. Hire when a clearly defined, recurring job is blocking progress and the founders cannot cover it, not because a funding round arrived.

Our article on first hires for a startup covers which roles typically come first, how to pay early employees with a mix of salary and equity, and how much runway to protect before adding fixed costs.

Common mistakes when starting a startup

• Building for months in secret. The longer you build without user contact, the more you build on untested assumptions.
• Targeting "everyone". A narrow first segment is easier to reach, easier to satisfy and easier to learn from.
• Treating compliments as validation. "Great idea" is not demand. A payment or a signed pilot is.
• Skipping the founder agreement. Unclear ownership is one of the most damaging and avoidable problems for young companies.
• Not tracking cash. Founders who do not know their runway tend to discover the problem when it is too late to fix.
• Raising too early or too late. Too early means poor terms; too late means negotiating with little runway left.
• Copying big-company processes. Early on, speed and direct customer contact beat elaborate planning documents.

A 90-day starting plan

Weeks	Focus	Output
1–2	Problem selection and first interviews	Problem statement, 10+ conversations
3–4	More interviews, solution sketches	Clear target segment, first mock-ups
5–6	Pre-sales or pilot offers	At least a few concrete commitments
7–10	Build the MVP	Working core workflow with first users
11–12	Measure, talk to users, decide	Go, pivot or stop decision based on evidence

Use the weekly review in our founder productivity system to keep this plan on track: one priority per week, a short list of experiments, and an honest look at what you learned.

The short version

Starting a startup is a sequence of bets, each one cheaper than the last if you learn fast. Choose a real problem, prove demand with commitments, build only what tests your key assumption, keep a close eye on runway, and bring in co-founders, investors and employees when the evidence says it is time. Everything else can be improved along the way.