Lighthouse vs. PageSpeed Insights: What's the Real Difference?

"I got 95 on Lighthouse but 62 on PageSpeed Insights — which one is right?"

This is one of the most common questions in web performance, and the answer is nuanced. Both tools use the same underlying engine, but they measure different things and serve different purposes.

Lighthouse: The Lab Tool

Lighthouse is an open-source auditing tool built by Google. It runs in a controlled, simulated environment (your machine or a CI server) and produces scores based on lab data.

What Lighthouse Does

• Loads your page in a simulated mobile environment
• Throttles CPU (4x slowdown) and network (simulated slow 4G)
• Measures performance metrics during a single page load
• Runs accessibility, SEO, and best practices audits
• Produces a deterministic (mostly) score

Where Lighthouse Runs

• Chrome DevTools (Lighthouse tab)
• Chrome Extension
• CLI: npx lighthouse https://yoursite.com
• CI/CD pipelines (Lighthouse CI)
• Node.js API

Lighthouse Strengths

• Consistent environment — same throttling every time
• Actionable diagnostics — tells you exactly what to fix
• Runs locally — can test staging/dev environments
• No traffic required — works on brand new pages

Lighthouse Limitations

• Not real user data — simulated throttling ≠ actual user experience
• Single-run variability — scores can fluctuate 5-10 points between runs
• Desktop bias — even with mobile simulation, it runs on powerful hardware
• No geographic variation — tests from one location only

PageSpeed Insights: The Hybrid Tool

PageSpeed Insights (PSI) is a web-based tool that combines Lighthouse lab data with Chrome User Experience Report (CrUX) field data.

What PSI Shows You

1. Field Data (top section) — Real metrics from actual Chrome users over the last 28 days
2. Lab Data (bottom section) — A fresh Lighthouse run performed by Google's servers
3. Diagnostics — Same actionable recommendations as Lighthouse

PSI Strengths

• Real user data — shows what actual visitors experience
• Geographic distribution — aggregates data from users worldwide
• SEO-relevant — CrUX data is what Google uses for ranking signals
• Device-specific — separate mobile and desktop field data

PSI Limitations

• Requires traffic — CrUX data only available for sites with enough Chrome visitors
• 28-day rolling average — slow to reflect recent improvements
• Origin-level data — for low-traffic pages, shows origin (whole site) data instead
• No staging access — can only test publicly accessible URLs

Why Scores Differ

Different Environments

Lighthouse in DevTools runs on your machine with simulated throttling. PSI's lab data runs on Google's servers. Different hardware = different results.

Field vs. Lab

PSI's overall assessment (Good/Needs Improvement/Poor) comes from field data, not lab data. Your Lighthouse lab score could be 95 while your field CLS is 0.3 (Poor) because real users on slow devices trigger layout shifts your lab test doesn't.

Metric Weighting

Lighthouse weights metrics differently than field data thresholds:

Metric	Lighthouse Weight	Field Threshold (Good)
LCP	25%	≤ 2.5s
TBT/INP	30%	≤ 200ms
CLS	25%	≤ 0.1
FCP	10%	≤ 1.8s
Speed Index	10%	—

When to Use Which

Use Lighthouse When:

• Developing locally and testing changes
• Running automated tests in CI/CD
• Debugging specific performance issues
• Testing unreleased pages or features
• Comparing before/after optimizations

Use PageSpeed Insights When:

• Checking what Google "sees" for SEO purposes
• Verifying field data for Core Web Vitals ranking signals
• Auditing competitor pages
• Presenting performance data to stakeholders
• Confirming that lab improvements translated to field improvements

Use Both When:

• Performing a comprehensive performance audit
• Identifying gaps between lab and field performance
• Setting up monitoring baselines

The "Third Option": Continuous Monitoring

Both Lighthouse and PSI give you a snapshot. Performance changes with every deployment, every new third-party script, and every traffic pattern shift.

What you really need is continuous monitoring that tracks performance over time and alerts you when metrics degrade.

What to Monitor

• Run Lighthouse audits on a schedule (every 6 hours, daily)
• Track trends over time, not single scores
• Set threshold alerts (e.g., "alert me if LCP exceeds 3s")
• Monitor after every deployment

Common Misconceptions

"I need a 100/100 score"

No. A score of 90+ is "Good." Chasing 100 often means sacrificing functionality (removing analytics, stripping fonts) for marginal gains. Focus on the metrics that impact user experience and conversions.

"My competitor scores higher, so they rank higher"

Core Web Vitals are a tiebreaker signal, not a dominant ranking factor. Content relevance, backlinks, and domain authority still matter far more.

"I fixed the issue but my PSI score didn't change"

CrUX field data updates on a 28-day rolling basis. Your fix might take 2-4 weeks to fully reflect in field metrics.

Stop Checking Scores Manually

Running PageSpeed Insights manually is tedious and doesn't scale. BadPageSpeed automates Lighthouse audits across all your pages and tracks scores over time.

Set up automated monitoring →

Related Reading

• Core web vitals 2026 complete guide
• How to prioritize performance fixes impact matrix
• Speed optimization roadmap 30 60 90