End-to-End Testing

An end-to-end (E2E) test simulates a complete user journey through an application, from the user interface down to the back-end systems (databases, APIs, third-party services). It validates that all technical layers function correctly together under conditions close to production. Unlike unit tests that isolate a single component, an E2E test checks the overall behavior of the system. It is one of the pillars of a quality strategy to guarantee reliability before going live.

Horizontal E2E tests cover a cross-functional user journey that spans multiple applications or modules (e.g., placing an order on an e-commerce site, from product selection to payment completion). Vertical E2E tests validate a deep workflow through all layers of a single application (UI, business logic, database). Both approaches are complementary: horizontal testing validates the overall user experience, while vertical testing guarantees technical consistency. The choice depends on the functional context and criticality.

E2E tests detect integration defects that unit or integration tests cannot see, such as issues between services, data inconsistencies, and regressions on user journeys. They provide high confidence before each production deployment by simulating real usage. They are particularly critical for workflows with high business value (payments, registration, regulated processes). Without E2E tests, an application may function unit by unit but fail in its overall usage.

The lifecycle begins by identifying critical user journeys based on user stories or functional specifications. Next comes designing the test cases, implementing them (manually or through automation), running them in representative environments, and analyzing the results. Detected defects are reported and fixed, and then the tests are rerun to validate the fix. Finally, E2E tests continuously evolve alongside the application, which requires regular maintenance.

Manual E2E tests rely on a tester who executes the journey step-by-step, allowing for detailed observation but becoming costly and slow at scale. Automated E2E tests are scripted and executed by tools, guaranteeing speed, repeatability, and execution with every deployment. Automation is ideal for regression testing, whereas manual testing remains relevant for exploratory testing and user experience evaluation. The best strategy generally combines both approaches based on criticality.

Unit tests validate an isolated function, integration tests check the interaction between multiple components, and E2E tests cover the entire system. The testing pyramid recommends a majority of unit tests (fast, targeted), an intermediate layer of integration tests, and a smaller number of E2E tests (slower and more expensive, but essential). Each level has its value, and none replaces the others. A mature testing strategy balances all three according to project demands.

The main challenges include slow execution speeds, fragility when faced with interface changes, the complexity of setting up test environments, and high maintenance costs. E2E tests are also prone to "flakiness" (intermittent failures not related to actual bugs), which erodes team trust. Finally, they can be difficult to debug because they traverse numerous technical layers. Modern solutions like Thunders.ai address these challenges using AI and self-healing.

Popular open-source frameworks include Cypress, Playwright, Selenium, and Puppeteer for the web, and Appium for mobile. Commercial platforms like Thunders.ai, Mabl, Testim, or Functionize add AI capabilities, self-healing, and advanced reporting. The choice depends on the language used by the team, the technical stack, and the desired level of automation. For teams seeking rapid setup and generative AI, all-in-one platforms like Thunders.ai offer an efficient balance.

Best practices include: targeting critical user journeys first, maintaining a reasonable number of tests to limit execution costs, isolating test environments, managing test data deterministically, and investing in stability (avoiding flakiness). It is also recommended to execute E2E tests within the CI/CD pipeline, but after faster tests have run. Finally, collaboration between developers and testers on test design ensures the business relevance of the scenarios.

Integration involves triggering E2E test execution automatically at specific stages of the pipeline, generally after unit and integration tests. Results are published in reports accessible to the entire team, and quality gates can block a release in case of failure. To optimize duration, execution can be parallelized across multiple environments, and intelligent test selection can be applied based on code changes. The goal is to obtain fast feedback without sacrificing coverage.