How to Build a Minimum Viable Product (MVP)

To build a minimum viable product (MVP), write down the one assumption that would sink the business if it were wrong, then build the smallest thing that lets real customers prove or disprove it. Decide before launch what result counts as a pass, put it in front of a small group of the people you want to sell to, measure what they do rather than what they say, and then keep going, change course or stop.

This page is about building and measuring the first version. To test whether there's demand before you build anything, see how to validate a business idea. To pay for the work, see how to fund your startup, and to put the result into a plan, see how to write a business plan.

Name the assumption you're testing

Write one sentence that starts with We believe. For example: We believe small cleaning companies will pay each month for a tool that turns a finished job into an invoice the same day. If you have several assumptions, rank them by how badly the business fails if each one is wrong, and test the top one first.

The US Small Business Administration's guidance on market research (US, checked October 31, 2026) says: "Market research blends consumer behavior and economic trends to confirm and improve your business idea." Its list of questions starts with: "Demand: Is there a desire for your product or service?" An MVP is one way to answer that question with behavior instead of opinions.

Choose a low-cost test that shows real behavior

An MVP doesn't have to be software. Pick the version that answers your assumption quickly:

• Do it by hand. Do the work for a handful of customers yourself, and only automate the steps that keep repeating.
• A page that asks for a commitment. Describe the offer and ask people to book a call, join a list or pay a deposit.
• One working path. Build only the screens needed to finish one task from start to end.

Whichever you choose, the one task it promises has to work every time. Leave out settings, rare cases and polish until the core task proves itself.

Decide what a pass looks like before you launch

Pick two or three measures and write the target next to each one:

• First use: how many invited people finish the core task once.
• Coming back: how many do it again within a set number of days.
• Paying: how many pay, or commit to pay, at the price you plan to charge.

Add a deadline. Set the end date before you start, and compare the results with your targets on that day.

Charge something early

A price is part of the assumption. Use free users to check whether the product works, and paying customers to judge whether it's worth building. If you're not ready to charge, ask for a deposit or a signed letter of intent and treat that as the signal.

Launch to a small group and watch

Invite a group small enough that you can talk to each person, such as a few dozen people who match your target customer. Set each one up yourself the first time, note where they get stuck, and fix those spots before you add anything new. Write down what people actually say, word for word, so you can spot the same problem coming up more than once.

Worked example: did the test pass?

Sample figures, for illustration only. A founder invites 40 people from the target group and sets these targets before launch: half finish the core task, and at least 8 pay.

• 26 of 40 finish the core task: 26 ÷ 40 = 65%, above the 50% target.
• 9 of those 26 pay $29.00 a month: 9 × $29.00 = $261.00 a month, and 9 is above the target of 8.

Both targets are met, so the founder keeps going and plans the next assumption to test. Had only 4 people paid, the honest answer would be to change the offer or the price before building more. Set your own targets for your own test.

Keep going, change course or stop

When the deadline arrives, compare the results with the targets you wrote down, not with what you hoped. If the test passed, move to the next riskiest assumption. If it failed, change one thing, such as the customer group, the offer or the price, and test again. If several honest tests fail, consider stopping before you spend more on building it.

Where Billed fits

If your first version is a service you deliver by hand, you still need to bill for it and know what it costs you. Billed's time tracking page says: "Log time for each task as you work, and see how much time goes into each project." Billed's invoice software page says: "Attach receipts to your expenses, convert an approved estimate into an invoice, and bill for time you have tracked." If you test a monthly price, Billed's recurring invoices page says: "Negotiating fixed prices and sending recurring invoices makes it easier for you to forecast your income and upcoming workload."

Mistakes to avoid

• Testing several assumptions at once, so a result doesn't tell you which one failed.
• Setting the pass mark after you see the numbers.
• Giving the product away and calling sign-ups proof of demand.
• Adding features before the core task works for the first group.
• Running the test with friends instead of people who match your target customer.

When this guide isn't for you

If your product handles health data, money or anything else that's regulated, the first version has to meet those rules too. Talk to a lawyer before you put it in front of customers.

Related Articles

• How to Validate a Business Idea
• How to Fund Your Startup
• How to Write a Business Plan
• Business Tools for First-Time Founders

Frequently Asked Questions

What should an MVP test?

The one assumption that would sink the business if it were wrong, such as whether a specific group of customers will pay for the result. Test the riskiest assumption first and leave the others for later rounds.

How do I know if my MVP worked?

Write the targets and a deadline before launch, such as how many people finish the core task and how many pay. When the deadline arrives, compare the results with those targets, not with what you hoped to see.

Should I charge for an MVP?

Charging, or asking for a deposit, tests whether people value the result enough to pay. Free use can show whether the product works, but not whether there's a business in it.

Can I track the hours I spend running a manual MVP in Billed?

Billed's time tracking page says: "Log time for each task as you work, and see how much time goes into each project."