How to Start a SaaS Company

Who a SaaS company suits and what running one involves

SaaS stands for software as a service: customers pay a recurring fee to use software you host. It suits founders who can build the product themselves or have a technical partner, who understand one industry or job well, and who have the patience for slow, compounding growth. Each customer pays a little every month, so income builds gradually.

Day to day, a small SaaS founder writes less code than you might expect. A normal week includes sales calls and demos, answering support tickets, onboarding new accounts by hand, fixing bugs, watching sign-up and cancellation numbers, and deciding which requests to refuse. Once customers rely on the product, you are also on call when it breaks.

• Good fit: you have seen the same problem repeatedly at work, you can reach the people who have it, and you can fund several months of your own time.
• Poor fit: you want quick cash, you dislike selling, or you have no way to reach buyers except paid ads.
• Not technical? You can still start by proving demand, then find a co-founder or hire a developer with paying customers already lined up.

How to start a SaaS company in 10 steps

1. Start from a problem you have seen up close. Look for work that people do by hand every week in spreadsheets, email or copy and paste, or software they pay for and complain about. The B2B SaaS business ideas list is organized this way if you need a starting point.
2. Interview fifteen to twenty people who have the problem. Ask how they handle it now, how often it comes up, what it costs them in hours or money, and what they have already paid to fix it. Do not pitch. A problem nobody has tried to solve is usually not painful enough.
3. Size the market roughly. Count how many businesses or people fit your target and what they could pay per year. The TAM SAM SOM calculator turns that into a first estimate. You are checking that the niche can support a business, not building a forecast.
4. Ask for a commitment before you build. Options include a pre-order at a discount, a signed letter of intent, a paid pilot, or a "concierge" version where you deliver the result by hand with off-the-shelf tools. If nobody will commit, go back to the interviews.
5. Define the niche and the offer. Write one sentence: for whom, to do what, instead of which current method. A tool for "small businesses" is hard to sell. A tool that produces compliance reports for independent care homes is easy to explain.
6. Build the smallest useful version. A minimum viable product (MVP) for SaaS is one workflow done well, with login, payment and a way to contact you. Skip team permissions, integrations and settings pages until a paying customer needs them. How to build an MVP covers how to cut scope.
7. Charge from the first version. Pick a simple pricing unit that grows with the value customers get, such as per user, per location or per volume. Start with one or two plans. Raise prices for new customers as you learn. Free plans can come later, once you know what a paying customer looks like.
8. Plan the cost categories. See the cost section below, and keep fixed monthly commitments low until revenue covers them.
9. Win the first ten customers by hand. Go back to your interviewees and pilot users, ask each for one introduction, and join the communities where your buyers already talk. Onboard every account yourself so you see where people get stuck.
10. Formalize once revenue repeats. When customers renew and new ones arrive without your personal network, incorporate in the form your advisers recommend, put founder agreements in writing, and decide whether to keep growing from revenue or raise money.

You can work these out with the MRR calculator and the churn rate calculator.

Which way of testing demand should you use?

Each test below trades speed against strength of evidence. Use the lightest one that would change your mind, then move down the table.

What does it cost to start a SaaS company?

The costs depend heavily on who builds the product and how much data you process, so price each category for your own case. The startup cost calculator will total them.

• Building: your own time, a contractor or agency, or an employee. This is nearly always the largest item.
• Hosting and infrastructure: servers, database, file storage, backups, monitoring and email delivery. Most providers have free or low starting tiers, and the bill grows with usage.
• Third-party services: payment processing fees, any APIs your product calls (AI model usage can become a significant variable cost), and analytics or error tracking.
• Sales and marketing: a website, domain, demo tools, and any ads or events.
• Legal and admin: incorporation, terms of service and privacy policy, accounting, and insurance where customers ask for it.
• Security and compliance: larger customers may ask for security reviews or certifications. Ask early which ones your buyers need, because some take months.

Legal and compliance points to check

Software crosses borders, so you may have to consider the law where your customers are as well as where you are. A startup lawyer and an accountant who works with software companies are the right people to ask. Rules vary by country and state.

• Company form: which legal structure fits your plans, especially if you may raise investment later or have co-founders.
• Ownership of the code: written agreements assigning intellectual property to the company from every founder, contractor and employee. Check your current job contract if you are building on the side.
• Customer terms: terms of service, a privacy policy and, for business customers, a data processing agreement.
• Data protection: which privacy laws apply to the personal data you store and where it may be hosted. Health, finance and children's data often carry extra rules.
• Sales tax and VAT: digital services are taxed differently across countries and states. Ask when you need to register and collect.
• Open source licenses: check the license terms of the libraries you build on.

Common mistakes when starting a SaaS company

• Building for months before the first sales conversation. Most of what you learn about what to build comes from trying to sell it.
• Mistaking a waitlist for demand. Sign-ups and compliments are not payment.
• Pricing too low to afford selling. If the product is sold through demos and calls, a very low monthly price cannot cover the time each sale takes.
• Building what one loud customer asks for. Custom features for a single account turn a product into a consulting job.
• Ignoring cancellations. Every customer who leaves can tell you something. Ask them.
• Targeting a buyer you cannot reach. If you have no path to the person who signs the contract, a good product still will not sell.
• Adding a free plan too early. Free users create support work and tell you little about willingness to pay.

What to do this week

1. Write down three problems you have personally watched people struggle with at work, and pick the one that happens most often.
2. List twenty people who have that problem and message ten of them asking for a twenty-minute call about how they handle it.
3. After five calls, write the one-sentence offer and a rough monthly price.
4. Ask the two most interested people what it would take for them to commit to a paid pilot.
5. Do not open a code editor until someone has said yes.

Fonda follows the same order: its Validate phase covers the market, key risks and customer interviews before it gives a Verdict of go, refine or pivot, and its Launch phase includes an MVP plan. It is free to start, and the method is based on peer-reviewed entrepreneurship research: see the research behind Fonda.