Building resilient Playwright suites that don't flake

Flaky tests are worse than no tests. A suite that fails randomly trains a team to ignore red builds — and once people stop trusting the pipeline, automation stops paying for itself. Over years of building test automation, a few principles keep suites stable.

Wait for state, never for time

The single biggest source of flakiness is sleep. Playwright's auto-waiting and web-first assertions remove almost all of it — assert on the condition you care about and let Playwright poll.

// Don't: arbitrary timing, race conditions waiting to happen
await page.waitForTimeout(2000);
expect(await page.locator(".total").textContent()).toBe("€42.00");
 
// Do: assert on state, Playwright retries until it's true
await expect(page.getByTestId("cart-total")).toHaveText("€42.00");

Isolate every test

Shared state between tests is the second-biggest cause of intermittent failures. Each test should set up its own data and tear it down — no ordering assumptions.

• Seed data through the API, not the UI: it's faster and less brittle.
• Use a fresh browser context per test for clean cookies and storage.
• Never depend on a test that ran before you.

Make selectors intentional

Target elements by role or test id, not by brittle CSS paths that break on every restyle.

// Resilient: survives layout and styling changes
await page.getByRole("button", { name: "Continue" }).click();
await page.getByTestId("email").fill("user@example.com");

Run them like production

A suite that only passes locally isn't done. Run it in CI, in containers, in parallel — the same way it'll run for real. That's where the remaining flakiness hides, and it's exactly what a good Docker setup makes reproducible.

Stable automation is a feature. Treat flakiness as a bug with a root cause, not as background noise.