How do you maintain a large Selenium test suite and prevent test rot?

// WRONG — locators scattered in test methods
@Test
public void loginTest() {
    driver.findElement(By.id("username")).sendKeys("admin"); // hardcoded
}

// RIGHT — all locators in Page Object
public class LoginPage {
    private By usernameField = By.id("username");  // change once, fixes all tests
    private By passwordField = By.id("password");
    private By loginBtn      = By.cssSelector(".btn-login");

    public LoginPage typeUsername(String user) {
        driver.findElement(usernameField).sendKeys(user);
        return this;
    }
}

// Tag flaky tests explicitly — don''t hide them
@Test(groups = {"flaky"})
@Issue("JIRA-456 — intermittent timeout on slow CI")
public void checkoutTest() { ... }

// Exclude flaky from main pipeline
// Fix by end of sprint — otherwise delete the test

// Method name pattern: should{ExpectedBehavior}When{Condition}
public void shouldShowErrorWhenPasswordIsMissing() { }
public void shouldRedirectToDashboardAfterValidLogin() { }
public void shouldDisplayTotalPriceIncludingTax() { }

// Easy to understand failures without reading code

Week 1-3: Feature/test development
Week 4:   Maintenance week
  - Fix all flaky tests (or delete them)
  - Update deprecated locators
  - Remove obsolete tests (feature removed)
  - Add missing test coverage
  - Review test execution time trends

// Test has failed 5+ consecutive runs
// No one has fixed it in 2 sprints
// → DELETE the test, file bug separately

// Keeping broken tests is worse than no test
// They add noise and reduce trust in the suite

Flakiness rate:     Target < 2%  — Alert if > 5%
Pass rate trend:    Alert if drops > 3% week-over-week
Test count growth:  +5-10 new tests per sprint
Execution time:     Alert if suite grows > 30 min
Disabled tests:     Review all @Test(enabled=false) monthly