Manual testing vs automated testing: which does your product need?
When to test by hand, when to automate, and how most teams end up doing both. A practical guide with examples of what each catches.

Most teams ask this question the wrong way round. They ask which is better, when the useful question is which checks belong to people and which belong to a machine.
What manual testing is good at
A person notices things nobody wrote a rule for. The button works, but it is hidden behind the keyboard on a small phone. The form submits, but the error message blames the user for something they did not do. The page loads, but the price jumps a second later and looks like a trick.
Manual testing is the right choice when:
- The feature is new. Nobody knows yet what "correct" looks like, so there is nothing stable to automate.
- You are asking "does this make sense?" Layout, wording, confusing steps and anything a user would find odd.
- You need to explore. A good tester follows a hunch: "what if I press Back here?" That is where surprising bugs come from.
- It runs rarely. A check you do twice a year costs more to automate than to do.
What automated testing is good at
A machine does the same thing the same way every time, without getting bored. That makes it right for checks that are:
- Repeated on every release. Login, search, add to basket, checkout. If your team runs these by hand before each deploy, they are the first to automate.
- Stable. The journey has not changed in months and is not about to.
- Expensive to get wrong. If a broken checkout costs you sales every hour it goes unnoticed, you want a test that runs on every code change.
- Wide. The same journey on three browsers and four screen sizes is twelve runs. A person will skip some. A machine will not.
What each one misses
Automated tests only check what they were told to check. A test that confirms the "Pay" button exists will pass happily while the button sits off-screen. Automation tells you that something you already knew about has broken. It does not find new kinds of problem.
Manual testing does not scale. Testing every journey on every browser before every release takes days, so under deadline pressure it gets cut short, and the parts skipped are usually the ones that broke.
A simple way to decide
For each check, ask three questions:
- Will we run this more than ten times? If not, do it by hand.
- Is the feature stable? If it is still changing weekly, wait.
- Is the expected result exact? "The total is $107.10" can be automated. "The page feels cluttered" cannot.
Three yes answers means automate it. Anything else stays manual for now.
What most teams end up with
A healthy setup usually looks like this:
- Automated tests cover the five to fifteen journeys that must never break, and run on every code change.
- Manual testing covers each new feature before release, plus an exploratory pass on the whole product from time to time.
- Bugs found by hand become automated tests once fixed, so the same bug cannot quietly return.
That last habit matters most. It means your automated suite grows from real failures instead of guesses.
Where to start
If you have no testing at all, start manual. One careful pass over your main journeys will find more in a day than a week of writing tests, and it shows you which journeys are worth automating.
If you already test by hand before every release and it is slowing you down, automate the repeated part first.
We do both: manual website testing and mobile app testing for finding problems, and Playwright automation for keeping them fixed.