Make end-to-end (E2E) tests fewer, more independent, and tied to observable behavior. Keep them for critical paths and whole-system risks; cover smaller logic at lower test layers. Then isolate test state, choose resilient locators, wait for conditions instead of fixed delays, and treat retry-only passes as failures to investigate.
Keep the end-to-end layer purposeful
E2E tests exercise a user path across multiple parts of a running system. That makes them valuable for checking that important pieces work together, but typically slower and more expensive to maintain than smaller tests. Use them where the complete path or a system-level property matters; test smaller units and component interactions at lower layers when those can detect the same defect.
As an Amazon Associate I earn from qualifying purchases.
Google’s 2015 testing-pyramid guidance offers 70% unit, 20% integration, and 10% end-to-end as a starting heuristic, not a measured optimum or a universal prescription. Google’s later guidance likewise recommends reserving E2E tests for important use cases and behaviors that smaller tests cannot reliably assess. The right balance depends on the system and the defects the team needs to catch. Google’s testing-pyramid guidance and guidance on E2E tests explain the trade-offs.
Recommended Free Tools
- Keep an E2E test when it verifies a critical user journey or a behavior that depends on several integrated systems.
- Move checks to a smaller test layer when they concern local logic or a component interaction that layer can verify reliably.
- Review overlapping tests: several scripts repeating the same interaction may add runtime and repair work without covering a distinct risk.
As Adam Bender, a Google Testing on the Toilet contributor, notes: “An end-to-end test often necessitates multiple test doubles (fakes or stubs) for underlying dependencies; they can, however, have a high maintenance burden as they drift from the real implementations over time.” That is another reason to keep this layer focused.
Make each test independent
A test that depends on another test’s browser state, data, or execution order can fail for reasons unrelated to the behavior it is supposed to check. Playwright recommends giving each test its own storage, data, and cookies. Its documentation summarizes the benefit: “Test isolation improves reproducibility, makes debugging easier and prevents cascading test failures.” Cypress also recommends isolating specs and controlling application state; its E2E test isolation is enabled by default. See Playwright’s best practices, Cypress’s best practices, and Cypress’s guidance on organizing tests.
Control browser and application state
- Start each test with a known browser context rather than relying on cookies or local storage left by a previous test.
- Set up the required account, permissions, and records deliberately. Use disposable test data where practical so one run cannot contaminate another.
- Prefer programmatic setup, such as direct login, when the login interface is not the behavior under test. Keep a focused E2E test for the login flow itself if it is important.
- Make tests runnable alone as well as in the full suite. A test that passes only after another test has run has a hidden dependency.
When tests need shared resources
If a test environment cannot provide entirely separate resources, make the shared setup explicit and prevent concurrent runs from modifying the same records. Clean up disposable data where appropriate. A shared account or fixed record is a coupling to manage, not an assumption to leave implicit.
Choose selectors that survive UI work
Selectors based on styling classes, long CSS chains, or incidental DOM nesting can break when presentation changes, even if the user-visible behavior is unchanged. Prefer locators that reflect what a user can identify, such as a button by its accessible role and name. When semantic locators are unavailable or do not identify the target unambiguously, use a dedicated test attribute as an explicit application-to-test contract.
Free tools Windows power users keep installed
One-click scans. No signup required.
| Locator approach | What it communicates | Maintenance trade-off |
|---|---|---|
| User-facing role and name | The element’s accessible, user-visible meaning | Often aligns the test with how a person finds the control; the name or semantics may legitimately change with the interface. |
Dedicated test attribute, such as data-cy |
An explicit identifier reserved for tests | Separates the locator from styling changes, but the application team must preserve and maintain the contract. |
| Styling class or incidental DOM structure | Implementation or presentation detail | Can make tests sensitive to refactors that do not change the behavior being tested. |
Playwright advises prioritizing user-facing attributes and explicit contracts; Cypress recommends dedicated data attributes when selectors should remain separate from styling and behavior changes. Neither locator type is universally best: use the one that clearly identifies the intended target and can be maintained as the interface evolves. See Playwright’s locator guidance and Cypress’s selector guidance.
Wait for the expected state, not a timer
A fixed sleep assumes the application will always be ready after the same elapsed time. It may waste time on fast runs and still be too short on slow ones. Prefer framework actions that wait for their preconditions and assertions that wait for the desired outcome. For example, assert that a confirmation status becomes visible or that navigation reaches the expected URL rather than pausing for a guessed interval.
Playwright documents automatic actionability checks before actions and asynchronous assertions that wait for conditions to become true. These practices reduce timing races when the condition is the right one; they do not fix unstable environments, bad test data, or product defects. See Playwright’s test-writing documentation and its best practices.
Rank #4
- Wait for a meaningful signal: the target is visible, a status changes, or the expected URL appears.
- Use network or loading-state signals only when they correspond to the behavior under test; the application’s readiness is not always equivalent to network quiet.
- Keep a fixed delay only when elapsed time itself is the requirement, not as a substitute for observing readiness.
Treat retries as diagnostic evidence
Playwright retries are disabled by default. If enabled, a test that fails first and passes on retry is classified as flaky, not as a clean first-pass success. A retry can help expose intermittent failures or keep a pipeline moving while the team investigates, but it is not a durable repair. Cypress likewise warns that tests which retry every run consume time and become technical debt.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Track first-pass failures separately from final pipeline status so retries do not conceal suite health. For Playwright CI failures, the trace viewer can show a timeline, DOM snapshots, and network requests; its documentation describes configuring traces on the first retry. See Playwright’s retry documentation, its debugging guidance, and Cypress’s test-performance guidance.
Best Value
Investigate a retry-only pass
- Record that the first attempt failed even if a later attempt passed.
- Inspect the trace or other available diagnostics to identify what differed: state, timing, navigation, or a request.
- Reproduce the test independently and check its setup, shared data, and waits.
- Fix the underlying coupling or condition, then verify the test succeeds without relying on a retry.
Prioritize maintenance with suite data
Do not start by rewriting every test. Use the suite’s own signals to find the work most likely to improve reliability and execution time. Cypress recommends looking at the slowest tests and specs, tests that repeatedly retry, and UI elements whose interaction counts seem disproportionate to their importance.
- Slowest tests or specs: identify opportunities to remove redundant steps, split unrelated checks, or move lower-level assertions out of E2E.
- Repeated retry failures: investigate flakiness rather than treating a green final pipeline as proof of health.
- Repeated interactions: check whether many tests exercise the same low-risk control without adding distinct coverage.
These signals are prioritization aids, not automatic deletion rules. Preserve tests that protect critical paths or system-level behavior even when they take longer; simplify or remove coverage only when another appropriate test provides the needed protection. See Cypress’s test-performance guidance.
Or skip the browser setup
For screenshots used in visual checks, bug reports, or test diagnostics, ScreenshotNeo can return an image or PDF from one GET request instead of requiring you to set up a browser capture flow. It is a website screenshot API and MCP server for developers.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →ScreenshotNeo accepts cookie or consent banners before capture and removes more than 60 known consent platforms, newsletter popups, and chat widgets; each of those steps can be turned off. Bot checks or CAPTCHAs, blank pages, timeouts, failed loads, and cache hits are not billed, and the response identifies the page verdict and billing status. AI agents can use its MCP server tools, including take_screenshot, get_page_info, and capture_pdf. The free plan includes 1,000 screenshots a month with no card; paid plans start at $5 for 3,000. See the ScreenshotNeo documentation.
curl -G "https://api.screenshotneo.com/v1/shot" -d access_key=YOUR_API_KEY --data-urlencode url=https://stripe.com -o shot.webpReplace YOUR_API_KEY with your key and change the target URL as needed. The response is saved as shot.webp.
Sign up for ScreenshotNeo’s free plan to get 1,000 screenshots a month with no card.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →
