Using Headless Browsers to ship reliable and performant apps
summary
Uses Playwright (Chromium) and Github Actions to run synthetic browser sessions against each preview deploy URL and production, injecting the web vitals library to capture FCP and LCP, counting 2xx vs 3xx/4xx/5xx requests and console errors, snapping screenshots on failures, and optionally blocking Vercel deploys. Example run surfaces a missing image invisible in a screenshot; Checkly reports millions of headless checks per month and can re-run scripts every 10 minutes in production.
AI-written summary · about
chapters (11)
- 0:00Headless browsers for reliability
- 1:43Playwright with Next.js setup
- 4:17First Playwright test
- 4:41Reliability metrics in tests
- 5:57Error analysis and hidden issues
- 7:03Web vitals performance metrics
- 8:23GitHub Actions with Vercel
- 10:16Multi-step test flows
- 11:31Checkly blocking deploys
- 11:58Golden age of end-to-end testing
- 12:21Summary and resources
related talks
20mTesting Web Applications with Playwright
01h 13mThe multi-year quest for the best web test in the west
40mMonitoring as Code: E2E Testing in Production with Playwright
28mRethinking Browser Automation: Unleashing the Power of Puppeteer and Playwright
35mState of end-to-end testing for modern web applications
30mPlaywright: The Tool That Will Revolutionize End-to-End Testing
39mTesting Web Applications with Playwright
40mAutomated Performance Testing In Preproduction