End-to-End Test.
An end-to-end test (E2E test) checks an application the way a person uses it: an automated browser opens the website, clicks, fills in forms and verifies that the right thing happens at the end, such as a confirmation appearing or an order being saved. The whole path from screen to database is tested.
End-to-End Test — Explained in Detail
Software is tested at several levels. Unit tests check individual functions, for example whether a price calculation is correct. Integration tests check whether parts work together, such as a form and the database. End-to-end tests check the entire flow from the user's point of view: open the page, choose a product, add it to the basket, pay, see the confirmation. Widely used tools for this are Playwright and Cypress. They drive a real browser, also invisibly in the background, and report where a step fails.
The big advantage: E2E tests find bugs that other tests miss. A form can be technically correct and still not work because a cookie banner covers the submit button, a redirect is wrong or a script does not load on mobile. These are exactly the bugs that cost enquiries and revenue, and they often only surface when customers complain. An E2E test that runs through the most important paths before every release is insurance against such surprises.
The price is effort and maintenance. E2E tests run more slowly than unit tests and are more fragile: if a label changes or a page takes a second longer, a test can fail even though nothing is broken. Such 'flaky' tests undermine trust. The test pyramid has therefore proven itself: many fast unit tests, some integration tests and a few carefully chosen E2E tests for the paths that really matter. Good tests address elements by their role or a fixed test attribute, not by colours or positions.
With coding agents, E2E tests gain importance. When an AI writes code, you need an independent check that the application still works afterwards. At the same time, agents can write tests and even operate a browser themselves to check a page. Frameworks are appearing on GitHub in which AI agents run through flows like a testing department. This does not replace fixed tests, but it can help find bugs nobody thought of.
Which flows deserve an E2E test
Not every page needs an E2E test. Worthwhile are the paths on which money or enquiries depend: contact and quote forms, appointment booking, login, basket and payment, registration and important calculators. Add flows that often break, for example after library updates. For each of these paths, one test for the normal case and one for the most important error case, such as an invalid field, is usually enough.
E2E tests work best when they run automatically before every release, in a staging environment with test data. They complement usability testing with real people but do not replace it: a test checks whether something works, not whether it is understandable. How we safeguard web apps with tests is described on web app development and website maintenance.
Example: a cleaning company and its quote form
A constructed example: a cleaning company in Basel receives most of its jobs through a multi-step quote form. After an update, nobody notices for weeks that the last step no longer submits on iPhones. Only when enquiries drop noticeably does anyone look for the cause.
An E2E test is then set up: an automated browser fills in the form in a desktop and a mobile view, submits it and checks that the confirmation appears and the enquiry arrives in the test database. The test runs before every change. If it fails, nothing goes live until the bug is fixed.
Related Page
Staging environmentFrequently Asked Questions About End-to-End Test
A unit test checks a single function in isolation and runs in milliseconds. An end-to-end test checks an entire flow in a real browser, from the interface to the database. It is slower and more effort but finds bugs in how the parts interact.
Playwright and Cypress are widely used, both open source. They drive real browsers, including mobile views, and can be integrated into automated pipelines before every release. In addition, tools are emerging in which AI agents check flows on their own.
Rarely for a purely informational site. As soon as a form, a booking or a shop brings in enquiries and revenue, at least one test for that path pays off. An unnoticed broken form usually costs more than the effort for the test.
More Terms Starting with "E"
E-E-A-T
E-E-A-T stands for Experience, Expertise, Authoritativeness, Trustworthiness — Google's quality fram...
Edge Computing
Edge Computing moves data processing from the central server to the 'edge' of the network — closer t...
Entity SEO
Entity SEO optimises for clearly defined entities — brands, people, places, concepts — and the relat...
Ready for Your Project?
Apply this knowledge to your website — DLM Digital will help you.