Software Testing

Beginner path · lesson 33 of 35

BeginnerAutomation5 min readUpdated November 12, 2026

Manual vs automated testing: when to use each

Manual and automated testing aren't rivals. Each is good at different things. Learn what each does best, what automated tests look like, how to decide which one a check needs and what it means for your career.

On this page

Key takeaways

  • Manual testing is a person using the software and judging the result.
  • Automated testing is code that runs checks and compares results with what's expected.
  • Automate checks you run often and that rarely change. Keep people on what needs judgment.

Manual testing

A person uses the software as a real user would: clicking, typing, reading and noticing.

Its strength is human judgment. A person spots a confusing label, a layout that looks broken or a flow that feels slow, even if nobody wrote a check for it.

Its weakness is repetition. Running the same 200 checks by hand before every release is slow, boring and error-prone.

Automated testing

A script drives the software and checks the result. For example: "Log in with a valid user, then check that the dashboard shows the user's name."

Its strength is speed and consistency. A suite of hundreds of tests can run on every code change, in minutes, the same way every time.

Its weakness is that it only checks what it was told to check. It won't notice that the button is now bright pink.

What an automated test looks like

Here's a short UI test written with Playwright, a popular browser automation tool:

test("a registered user can log in", async ({ page }) => {
  await page.goto("https://shop.example.com/login");
  await page.getByLabel("Email").fill("ana@example.com");
  await page.getByLabel("Password").fill("correct-horse-42");
  await page.getByRole("button", { name: "Log in" }).click();
  await expect(page.getByText("Welcome, Ana")).toBeVisible();
});

It does exactly what a manual tester would do, then asserts the expected result. If the welcome text doesn't appear, the test fails and tells you so.

Not all automated tests click through a browser

KindWhat it checksSpeed
Unit testsOne small piece of code, e.g. a function that calculates VATMilliseconds
API testsThe server directly, without the screenFast
UI / end-to-end testsThe whole app through the browser or phone, like a userSlowest

Most of a healthy suite sits in the fast, lower layers. The test pyramid explains why.

Side by side

ManualAutomated
Good atExploring, usability, one-off checksRepetitive checks, regression, data-heavy cases
SpeedSlow per runFast per run
Up-front costLowHigher (someone writes and maintains code)
Notices the unexpectedYesNo, only what it asserts

How to decide

Ask three questions about a check:

  1. How often will we run it? Every release or every commit points to automation.
  2. How stable is the feature? If the screen changes every week, the test will need rewriting every week.
  3. Does it need human judgment? "Does this look right?" or "Is this easy to use?" stays manual.

Do

Automate the login, checkout and other critical flows that must work on every release.

Don't

Automate a feature still being designed. You'll spend more time fixing the test than testing.

Real-world exampleThe regression trap

A small online shop tested everything by hand. Each release took the one tester two full days of the same checks.

To save time, they started skipping "old" areas. A change to discount codes then broke the checkout for customers paying in another currency. Nobody noticed for a week.

They automated the 30 most important checkout scenarios. The suite ran on every change in under 10 minutes, and the tester went back to exploring new features.

Lesson: automate what is repetitive and critical, so people can focus where judgment matters.

Common myths

  • "Automation will replace testers." Automation replaces repeating checks, not thinking about what to check. Someone still has to decide what matters, design the tests and investigate failures.
  • "We should automate 100% of our tests." Some checks are cheaper by hand, some change too often, and some need judgment. Aim for the right checks, not a percentage.
  • "Automated tests find lots of new bugs." Mostly they catch regressions: things that used to work and broke. New bugs are more often found by people exploring.
  • "Manual testing is unskilled." Good manual testing means designing sharp test cases, exploring systematically and reporting clearly. Those are the skills automation is built on.

The cost over time

An automated test costs more the first time, because someone has to write it, and a little every time the feature changes. Then each run is almost free.

A manual check costs about the same every time. So the more often a check runs, the sooner automation pays off. A check you'll run once a year rarely justifies the code. A check that runs on every code change, dozens of times a day, almost always does.

What this means for your career

Many testers start with manual testing: it teaches you the product, the domain and how to think about risk. Job ads then split into Manual QA and QA Automation roles, and automation roles usually ask for programming skills.

You don't have to choose forever on day one. A good path is to get strong at manual testing first, then learn the basics of code and write your first tests. See QA job titles explained and writing your first automated test.

Check yourself

You need to check whether a new onboarding screen is easy to understand for first-time users. Manual or automated?

Show answer

Manual. "Easy to understand" is a judgment only a person can make. Once the screen is stable, you could automate a check that it loads and its buttons work, but not whether it's clear.