Levels of testing: unit, integration, system, acceptance

Why testing has levels

Software is built in layers, so testing checks it in layers too. Each level has a different scope, a different owner, and a different kind of bug it is best at catching. Skipping a level does not remove the risk, it just pushes the bug to a later and more expensive stage.

Read the levels from the inside out: one function, then functions working together, then the whole system, then the user's verdict.

Unit and integration testing

Unit testing checks the smallest testable piece of code, usually a single function or method, in isolation. Developers own it, and it runs fast. A unit test confirms that a discount function returns the right total for a given input.

Integration testing checks that separate units work together. The units may each pass alone but still break when they share data. This is where API and contract testing does heavy lifting, confirming that one module sends exactly what another expects.

• Unit: is this function correct on its own?
• Integration: do these modules agree at the boundary?

Most painful bugs do not live inside a single unit. They live in the gaps between units, which is why integration testing earns its keep.

System and acceptance testing

System testing checks the fully assembled product against the requirements. It tests end-to-end flows the way a user would move through them, including performance and error handling. This is broad functional work, often supported by manual testing and automated end-to-end suites.

Acceptance testing is the final level. Here the question is not does it work, but does it meet the user's needs and the business rules. User acceptance testing puts real or representative users in front of the product, which is the heart of our exploratory and UAT sessions.

A concrete example

Picture an online store adding a wishlist. A unit test checks the function that adds one item to a list. An integration test checks that the wishlist service and the product catalogue exchange the right item ID. A system test walks the full journey: log in, browse, add to wishlist, move to cart, check out. An acceptance test asks a real shopper whether the wishlist actually helps them buy later.

• Unit and integration catch code and interface bugs early.
• System catches broken journeys across the whole product.
• Acceptance confirms the product solves the real problem.
• Wire the fast levels into CI/CD test integration so they run on every change.

One honest caveat: more levels do not mean more confidence if the tests are shallow. A green pipeline with weak assertions can hide real defects, so the quality of each test matters more than the count. Tools like Playwright and Selenium help at the system level, while the ISTQB syllabus defines each level in detail.

At Quality AboveAll, senior testers balance the four levels to fit your risk and budget, often for less than one full-time hire. If you want a clear plan for where to invest, book a free 30-minute testing audit.