Update E2E testing guidelines to minimize test count

Clarify that agents should create one test per user journey rather than separate tests for each validation, accessibility check, or Welsh translation. Added concrete examples showing good and bad patterns to prevent test proliferation.

Key changes:
- Emphasize "one test per user journey" principle
- Show comparison between minimal journey tests vs excessive granular tests
- Clarify that validations/accessibility/Welsh checks belong inline within journey tests

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

CLAUDE.md

@@ -366,17 +366,64 @@ describe('UserService', () => {
 
 ### E2E Testing with Playwright
 
+**CRITICAL: Minimize Test Count**
+
+**Create the minimum number of tests.** Each test should represent a complete end-to-end user journey. Do NOT create separate tests for individual validations, accessibility checks, or Welsh translations. Instead, include all validations, accessibility checks, and Welsh translations within the journey test itself.
+
+**Key Principle:** One test per user journey, not one test per validation or feature.
+
+✅ **Good - Minimum tests, each covering complete journey:**
+```typescript
+// One test for the subscription journey - includes validations, Welsh, accessibility
+test('user can subscribe to updates @nightly', async ({ page }) => {
+  await page.goto('/subscribe');
+
+  // Test validation along the journey
+  await page.getByRole('button', { name: 'Continue' }).click();
+  await expect(page.getByText('Enter your email')).toBeVisible();
+
+  // Test Welsh translation along the journey
+  await page.getByRole('link', { name: 'Cymraeg' }).click();
+  await expect(page.getByText('Rhowch eich e-bost')).toBeVisible();
+
+  // Test accessibility along the journey
+  const results = await new AxeBuilder({ page }).analyze();
+  expect(results.violations).toEqual([]);
+
+  // Complete the journey
+  await page.getByLabel('Email').fill('test@example.com');
+  await page.getByRole('button', { name: 'Continue' }).click();
+  await expect(page.getByText('Subscription confirmed')).toBeVisible();
+});
+
+// Separate test only for a DIFFERENT user journey (e.g., unsubscribe)
+test('user can unsubscribe @nightly', async ({ page }) => {
+  await page.goto('/unsubscribe');
+  // ... complete unsubscribe journey with validations, Welsh, accessibility
+});
+```
+
+❌ **Bad - Too many separate tests:**
+```typescript
+test('shows validation error for email', async ({ page }) => { /* ... */ });
+test('shows validation error for name', async ({ page }) => { /* ... */ });
+test('Welsh translation works on subscribe page', async ({ page }) => { /* ... */ });
+test('accessibility passes on subscribe page', async ({ page }) => { /* ... */ });
+test('user completes subscription', async ({ page }) => { /* ... */ });
+```
+
 **Test Organization:**
 - Location: `e2e-tests/`
 - Naming: `*.spec.ts`
 - Tag nightly-only tests with `@nightly` in test title
 
-**Test Patterns (Sequential in Each Test):**
-1. Test content and functionality
-2. Test Welsh translation (same journey)
-3. Test accessibility inline (not separate)
-4. Test keyboard navigation
-5. Test responsive behavior
+**What to Include in Each Journey Test:**
+1. Complete user journey from start to finish
+2. All relevant validation checks encountered in the journey
+3. Welsh translation checks at key points
+4. Accessibility checks at key points
+5. Keyboard navigation where relevant
+6. Successful completion of the journey
 
 **Example Pattern:**
 ```typescript
@@ -420,7 +467,6 @@ test('user can complete journey @nightly', async ({ page }) => {
 - Clean up test data in global-teardown.ts
 
 **Coverage Expectations:**
-- Business logic: >80%
 - E2E tests: Cover critical user journeys
 - Accessibility: Test inline with journeys (not separately)