How much testing does an MVP need?
A first version does not need a full QA process, but it does need some testing. Here is how to decide what to test, what to leave, and when to add more.

Founders usually hear two kinds of advice about testing a first version. One says to ship fast and fix things later. The other says an untested product will destroy your reputation. Both are right about something, which is why neither is much help.
A more useful question is which failures you can afford. Some bugs are embarrassing and some are fatal, and an MVP only has time for the second kind.
Sort your features by what breaks if they fail
Take every feature in your first version and put it in one of three groups.
Group 1: failure loses money, data or trust. Sign-up and sign-in. Payments. Anything that stores or shows personal information. The one action your product exists for, whether that is booking, ordering, sending or publishing.
Group 2: failure annoys people. Search and filters, profile editing, notifications, settings, dashboards.
Group 3: failure is barely noticed. The About page, animations, rarely used options, layout details at unusual screen sizes.
For an MVP, test Group 1 properly, give Group 2 one honest pass, and leave Group 3 for later. Most teams do the opposite. They polish what is visible and assume the payment flow works because it worked once.
What "properly" means for Group 1
Testing a feature properly does not mean checking that it works. It means checking what happens when it is used wrongly.
For each Group 1 feature, cover four things:
- The normal path, start to finish, on a phone and on a desktop.
- Bad input: empty fields, wrong formats, very long text, the same button pressed twice.
- Interruptions: the Back button, a refresh, a closed tab, a lost connection.
- The wrong person: signed out, or signed in as a different user.
For a small product this is a day or two of work. It is also the testing that catches the bugs people write angry reviews about.
What you can safely leave
Being honest about this saves real time.
- Every browser ever made. Test current Chrome and Safari, on desktop and on a phone. Add others when your analytics show people using them.
- Automated tests for everything. At MVP stage your screens change every week, and tests written today are thrown away next month. Automate later, once the main flows stop moving.
- Load testing. If you have fifty users, you do not need to know what happens at fifty thousand. One exception: if you are about to be featured somewhere and expect a spike on a known date.
- Perfect layouts. A button two pixels out of line has never lost anyone a customer.
What you cannot leave
Two things are worth doing even on the smallest budget.
Have someone else use it. The person who built a product follows the path they built. A stranger will do something different within the first minute. One outside tester for half a day finds more than a week of the founder clicking around.
Check security basics. Can one user see another's data by changing a number in the address? Can a signed-out visitor open a private page? These take minutes to check and are the mistakes that end young companies. Our guide to security testing lists the checks.
When to add more
Testing should grow with the product. These are the usual signs that it is time.
- You have paying customers. A bug now costs money, and you should rerun the Group 1 checks before every release.
- You release every week. Repeating the same checks by hand gets slow, and people start skipping them. This is when automated tests pay off, starting with sign-in and payment only.
- A fixed bug comes back. That means changes are breaking old features and nothing is catching it.
- More than one person is changing the code, or an AI tool is writing a large share of it. See why apps built with AI need testing more than any other.
A plan for a typical first version
If you want something to copy, this fits most MVPs:
| When | What | Time |
|---|---|---|
| Before launch | Group 1 tested properly, Group 2 one pass, security basics | 2 to 3 days |
| Before each release | Group 1 normal paths, on phone and desktop | 1 to 2 hours |
| After 2 to 3 months | Automate sign-in and payment checks | A few days, once |
The times are rough and depend on the size of the product. The order is what matters: protect what loses money first, repeat it at every release, and automate only when the flows have settled.
Paying for it
A full-time tester is hard to justify for a first version, and a monthly retainer often is too. For an MVP it usually makes more sense to pay for a fixed piece of work, or per bug found, so the cost follows the amount of trouble in the product.
If you are close to launch and want the Group 1 list tested by someone who did not build it, start a project. Tell us what the product does and we will reply with what we would test first.