Not whether to automate but what Automating the wrong things is a common way to spend a lot of money and still ship bugs, so the order matters more than the tooling.
What manual testing is good at
Judgement. A person notices that the page is technically correct but confusing, that the error message is rude, that the layout looks broken on their phone, and that a flow which passes every assertion still feels wrong. Exploratory testing, where a tester follows their own hunches instead of a script, finds the bugs a checklist was never written for. New features, one-off releases and anything visual belong here.
What automation is good at
Repetition. A machine will run the same two hundred checks at three in the morning, across several browsers, without getting bored or skipping step four. Automation is how you know that today's change did not break something you finished three months ago, and it is the only practical way to test a large app on every release.
What automation actually costs
Writing a test is usually the cheap part. The real cost is maintenance: selectors change, flows change, and a suite nobody maintains becomes a suite nobody trusts. Budget for ongoing upkeep from the start, and be honest that a test which fails randomly is worse than no test, because it teaches the team to ignore red builds.
A split that works
- Automate: login, search, cart and checkout, core product actions, API contracts, and the bugs you have already fixed once.
- Keep manual: brand-new features, visual design, content and tone, complex judgement calls, and exploratory sessions before a big release.
- Do both: accessibility, where a scanner runs on every build and a person does the keyboard and screen-reader pass.
The order to do it in
Run a manual cycle first. It tells you where the real risk is and which flows are stable enough to automate. Then automate those flows and let the manual effort move to whatever is new. That way automation always covers the boring, valuable part, and people always work on the part that needs a person.
Apex Automation Team does both: a manual pass on your critical journeys, then an automated regression suite you own, wired into your CI. See software testing and QA or ask for a test plan.
Quick checklist
- Run one manual cycle before automating anything
- Automate repeated, stable, high-value flows
- Keep new features and visual work manual
- Treat a flaky test as a broken test and fix or delete it
- Add a regression test for every bug you fix
- Review the suite quarterly and remove what no longer earns its place
Frequently asked
Is automation cheaper than manual testing?
Over time, for repeated checks, yes. For a one-off release or a brand-new feature, a manual pass is usually faster and finds more.
Can you do just manual testing?
Yes. Many clients start with manual cycles only, and automate later once the product has settled.