Guide · updated 16 November 2026
QA for startups: how to test without a QA team
A startup doesn't need a QA department. It needs one list of what the app must do, someone other than the author checking it, and a retest of every fix. This is how to set that up in a week.
Building an app for a client and building a product you want to own look the same at the beginning. They are not. For a client you get the requirements, build in phases, test it, hand over, and support it for as long as you agreed. With your own product you build only the core thing people should care about, put it in front of them, and then keep tweaking and improving it from their feedback. There is no handover. The app changes every week.
That is where a startup's testing falls apart. Most people don't test their app. They go through it and hopefully catch whatever they can catch on the way. We had the same issue with our own apps. It works once. It does not work when the app changes every week and the same three people are building, selling and supporting it.
1. What QA means at a startup
QA is not a team or a tool. It is three habits: a written list of what the app must do, a recorded result for each item on each platform, and a retest whenever something is fixed. A team of two can keep all three. A team of twenty without them still ships broken sign-up pages.
2. Write one checklist, from the code
List every journey in the app, one line each: what has to be true first, the steps, and the result a user should see. Group them by area (sign-up, the main job, payments, settings, admin). If you only list what you remember building, the parts you forgot are the ones that break.
If an AI tool wrote most of the code, have it write the list from the code. Our free test-checklist skill does this in Claude Code, Cursor or Codex (see how to test code written by Claude Code). For the format, start from the test plan template, and see how to write test cases for what a good line looks like.
3. Decide who tests
The person who wrote a feature is the worst person to test it, because they follow the path they built. Give each area of the checklist to someone who didn't write it: another developer, the founder, a designer, whoever answers support. They don't need to read code. They need the steps and the expected result, and a place to record what they saw.
Before a launch, also hand the app to two or three people who have never seen it and watch without helping. How to test your app before launch covers that week in order.
4. Test in this order
Sign-up, sign-in and password reset first. Then the one job the app exists to do. Then payments and every email the app sends. Then everything else. Do each on every platform the app runs on, and record a result per platform: a fix that works in the browser often never reached the phone.
Then test the paths nobody planned: empty and wrong input, Back and Refresh mid-way, a double tap on Pay, no network, a normal user typing an admin URL. The website testing checklist and the mobile app testing checklist list these.
5. Raise every bug against a check, and retest every fix
A bug reported in chat gets lost. Raise it against the line of the checklist it broke, with the platform, the steps and a screenshot (the bug report template has the fields). When a developer says it is fixed, the tester repeats the same steps on the same platform. Only then is it closed. Then check the screens around it, because fixes break neighbours: retesting and regression testing are different jobs and you need both.
6. Before each release, run the short list
You will not run the whole checklist every week. Run what changed, what was fixed, and the journeys that must never break: sign-up, the main job, payment. The release checklist is that short pass, and the regression testing checklist is the longer one for bigger releases.
7. Automate what you repeat, not everything
The journeys you run on every release are worth automating with Playwright or Cypress on the web and Maestro on phones. Leave new and fast-changing screens to a person until they settle. Write the automated tests from the same checklist, so a passing test and a passing check mean the same thing.
8. When to hire your first QA person
Hire when testing a release takes more than a day or two of the team's time, when you ship on several platforms every week, or when bugs reaching customers are costing you deals. Hire someone who will own the checklist and the automated tests, not only click through screens. An outside testing agency can help before a big launch, but it tests from the list you give it. Without a list you pay for the same guessing.
Keep it in one place
In qarunbook the checklist becomes a board with a result per platform. Testers record what they see and raise issues against a check; developers, or Claude Code over our QA MCP server, fix them and send them back for a retest. Anyone on the team can see how many checks have passed on each platform, so “can we release?” has a number for an answer. One person and one app are free. If you are comparing tools, see the best test management tool for a small team.