MVP development means building the smallest working version of a product that does one job well enough for real people to use and pay for, so you learn what is worth building next before you spend on the rest. An MVP (minimum viable product) is not a rough draft of the finished thing. It is a deliberately small product with a short list of features and a clear test.
How much it costs and how long it takes depends mostly on how short you keep that list. This article explains how to choose what goes in, what to cut, how to estimate the cost with your own numbers, and a first-week plan. We do not quote prices, because they vary by team, country and scope; we show you how to work out a sensible budget instead.
What an MVP is, and what it is not
The term comes from start-up practice. The idea is simple: instead of building everything you imagine, build the one core thing, put it in front of real users and see whether they use it and whether they would pay.
It is not:
- A prototype. A prototype shows how something could look or work. An MVP actually does the job, even if parts are manual behind the scenes.
- A cheap, poor version. It should be small and solid. A broken product teaches you only that it is broken.
- A product for everyone. It serves one type of user and one main task.
For an established small business, an MVP is often an internal tool or customer portal. The test is whether your staff or clients use it instead of the spreadsheet, email thread or phone call it replaces. Our guide to replacing a spreadsheet with an app covers a common case.
Choose the one job: how MVP development starts
Start with a single sentence in this shape:
[Type of person] can [do one thing] without [the current pain].
For example: “A site supervisor can log a variation with a photo from the job without phoning the office.” Or: “A new patient can complete their intake forms before the first visit without printing anything.”
Then test the sentence:
- Is there a real person you can name who has this problem today?
- What do they do about it now, and what does that cost them in time or money?
- Would they use a rough version this month if you handed it over?
- How would you know it worked? Pick a number: tasks done, hours saved, enquiries handled.
If you cannot answer the first question with a real name, stop. Building before you can name the user is how projects drift. See why software projects fail.
The cut list: what to leave out of the first version
The cut list is more useful than the feature list. Write down everything you want, then move items into three columns.
| Keep for the MVP | Do by hand for now | Leave for later |
|---|---|---|
| The one core task, start to finish | Admin screens: edit records in the database or a spreadsheet instead | Dashboards and detailed reporting |
| Login for the user type you serve | Approvals and notifications: someone phones or emails | Several user roles and permissions |
| Secure storage of the data it needs | Payments: send an invoice, collect later | Integrations with other software |
| A way to see it working on a phone | Onboarding and help pages: you walk the first users through | Native mobile app, fine design, extras |
“Do by hand” is the most important column. Many early products are a simple screen at the front and a person doing the rest behind it. That is fine. It is how you find out which parts are worth automating. It is also why starting as a web app rather than a mobile app suits most MVPs: one version, quick to change.
How much does an MVP cost? A way to estimate it
Search results for the cost of an MVP are full of figures that vary enormously. Any single number depends on the country, the team and the features, so none is quoted here. Work it out for yourself.
Software cost is mostly people’s time. Use this arithmetic:
- List the features you kept. Give each a size: small (a few days), medium (a week or two) or large (several weeks). Ask your builder to do this, not yourself.
- Add the days up. Add roughly a quarter more for design, testing, fixing and project management, which is where the surprise bills hide.
- Multiply by the daily rate of whoever builds it. The rate varies a lot by country and by team, so get quotes.
- Add running costs: hosting, email and text services, and someone to fix and update it. Our guide on software maintenance cost goes through this.
A worked example with invented numbers. Say your cut list leaves 6 small features and 2 medium ones. Small ones average 3 days and medium ones 8, so 18 + 16 = 34 days. Add a quarter (about 9 days) and you have roughly 43 days. If a builder charges $X a day, the build is 43 x $X.
Compare that with what the problem costs you. If staff lose 10 hours a week to the old way and an hour is worth $50 to the business, that is $500 a week or $26,000 a year. A build that costs a fraction of that and pays back inside a year is worth discussing. One that costs several years of the loss is not.
For a wider view of budgets, see app development cost.
How long MVP development takes
The same arithmetic gives the timeline. 43 working days is about nine weeks of one full-time builder, and less with two people working in parallel, though not half. Time is also lost waiting for you: decisions, content, logins and test feedback. A realistic plan is built around your side of the work as much as the builder’s.
We do not promise a timeframe in advance. A fair builder will give you a range after seeing your cut list, set milestones you can see working every one to two weeks, and tell you early when something is growing.
- First two weeks: agree the one-sentence job, the cut list and the screens on paper.
- Middle stretch: build the core task first, then login and data, then the rest.
- Last stage: real users try it with real work, and you fix what they stumble over.
Get real users in early
The point of an MVP is learning. Put it in front of three to five real users as soon as the core task works, even if it looks rough. Watch them use it. Do not explain; see where they hesitate.
Ask them three questions afterwards: what did you expect to happen, where did you get stuck, and would you use this next week instead of what you do now? Write the answers down, because memory drifts towards what you hoped to hear.
Then decide with your numbers. If people use it and the measure you chose moves, build the next item from the “later” column. If they do not, change the job or stop. Stopping after a small spend is a result and not a failure.
Own what you build
An MVP can turn into the real product, so protect your position from the start. Make sure the contract says who owns the code, the data and the accounts it runs on, and that you can get a copy of the code. We cover this in who owns the code. Ask for a short written description of what is being built; our guide on writing a software requirements document shows how.
CentriWeb is a founder-led technology studio based in New Zealand and working online with businesses worldwide. We build custom software such as portals, web apps and internal tools, starting with the part that costs the business most, and we look after it after launch. If an off-the-shelf product covers your one job, we will say so.
Your first week
- Day 1. Write the one-sentence job and name the first real user.
- Day 2. Interview that user for 20 minutes. Ask what they do now, how long it takes and what goes wrong.
- Day 3. Write the cut list in three columns. Be strict.
- Day 4. Pick the number that will tell you it worked, and note what it is today.
- Day 5. Check whether a ready-made tool already does the job. If it does, use it and save the build.
- Next week. Take the cut list to two builders and ask for sizes, an estimate, a range of time and who owns the result.