Slow sites lose users. But "performance" isn't one thing — it's lab testing, field monitoring, error tracking, and synthetic checks. Lighthouse, WebPageTest, Sentry, and Checkly each cover different parts of the performance puzzle. Here's how to build a complete monitoring stack.
Quick Comparison
| Lighthouse | WebPageTest | Sentry | Checkly | |
|---|---|---|---|---|
| Type | Lab testing (simulated) | Lab testing (real devices) | Error tracking + RUM | Synthetic monitoring + E2E |
| Best for | Quick audits, CI integration | Deep performance analysis | Catching production errors | Uptime + performance SLAs |
| Data source | Simulated throttling | Real devices, real networks | Real user sessions | Synthetic (global locations) |
| Core Web Vitals | Yes (lab only) | Yes (lab + field) | Yes (RUM — real users) | Yes (synthetic) |
| Free tier | Free (open source) | Free (public tests) | Free (5K errors/mo) | Free (50K checks/mo) |
| CI/CD | Lighthouse CI | WebPageTest API | Release tracking | Playwright-based checks |
Lighthouse — The First Line of Defense
Lighthouse is built into Chrome DevTools and runs simulated audits for performance, accessibility, SEO, and best practices. Lighthouse CI lets you set performance budgets and fail builds that regress. It's the starting point for any performance effort.
Strengths: Free and built into Chrome. One-click audits. Lighthouse CI for build-time checks. Performance budgets in CI. Clear, actionable recommendations. Covers perf, a11y, SEO, and best practices in one report.
Weaknesses: Lab data only (simulated, not real users). Scores vary between runs. Simulated throttling doesn't match real-world conditions. Doesn't catch real-user issues that only appear in production. Single-device simulation.
Best for: Quick audits during development, CI performance budgets, catching regressions before deploy, the starting point for any performance optimization.
WebPageTest — The Deep Performance Debugger
WebPageTest runs your site on real devices with real network conditions in locations worldwide. The waterfall chart, filmstrip view, and connection-level details reveal exactly what's slowing your site down. If Lighthouse tells you "what" is slow, WebPageTest tells you "why."
Strengths: Real devices (Moto G4, iPhone, etc.) on real networks (3G, 4G). Waterfall chart shows every request. Filmstrip view shows visual progress. Multi-location testing. Advanced features (scripting, custom metrics). Free for public tests.
Weaknesses: Not for continuous monitoring (spot tests). More complex than Lighthouse. Free tier is public (your test results are visible). No real user monitoring.
Best for: Deep performance debugging, optimizing critical rendering path, comparing before/after optimizations, understanding real-device performance.
Sentry — Real User Error & Performance Monitoring
Sentry captures real errors and performance data from actual users. When your app crashes in production or a page takes 10 seconds for users in a specific region, Sentry tells you — with the stack trace, user session, and breadcrumbs to reproduce it.
Strengths: Real user errors with full context (stack trace, user, session replay). Performance monitoring (slowest routes, DB queries, API calls). Release tracking (did the deploy cause a spike?). Session replay for debugging. Open source (can self-host). Excellent SDKs (30+ languages).
Weaknesses: Can be expensive at scale (many errors/transactions). Noise requires tuning (alert fatigue). Not for synthetic monitoring. Session replay costs extra. Self-hosting requires maintenance.
Best for: Production error tracking, identifying slow transactions for real users, catching regressions after deploys, debugging user-reported issues.
Checkly — Synthetic Monitoring for Production
Checkly runs Playwright-based browser checks from 20+ global locations on a schedule. It verifies that your key flows work — login, checkout, search — and alerts you when they don't. It combines API checks, browser E2E checks, and performance monitoring in one platform.
Strengths: Playwright-based (real browser checks). Global monitoring (20+ locations). API + browser checks combined. Performance trending over time. Alerting (Slack, PagerDuty, email). Terraform/CI/CD integration. Generous free tier. Status pages built-in.
Weaknesses: Synthetic only (not real users). Setup requires writing Playwright scripts. Free tier limited to 50K check runs/month. Less useful for SPA-heavy apps without careful scripting.
Best for: Uptime and performance monitoring, SLA compliance, testing critical user flows in production, catching issues before users report them.
Building a Complete Monitoring Stack
| Layer | Tool | Frequency |
|---|---|---|
| Dev-time audit | Lighthouse (Chrome DevTools) | Every PR |
| CI performance budget | Lighthouse CI | Every build |
| Deep performance debug | WebPageTest | Before/after optimizations |
| Real user errors + perf | Sentry | Continuous (production) |
| Synthetic monitoring | Checkly | Every 5-15 min (production) |
Bottom line: Lighthouse for dev and CI, Sentry for production errors and real-user performance, Checkly for synthetic uptime/flow monitoring, WebPageTest for deep dives. These four tools together cost $0 for small projects and give you complete visibility. See our hosting comparison — good hosting makes performance easier.