Why 100/100 on Lighthouse Isn't Always the Goal

The pursuit of a perfect 100/100 Lighthouse score has become an obsession for some development teams. But here's the uncomfortable truth: a perfect score often comes at the cost of functionality, features, and even revenue.

The Problem with Perfection

You'll Remove Useful Third-Party Tools

To hit 100, you'd likely need to remove:

• Analytics (Google Analytics, Mixpanel) — losing conversion tracking
• Chat widgets (Intercom, Drift) — losing customer support
• A/B testing tools (Optimizely, VWO) — losing optimization capability
• Retargeting pixels (Meta, Google Ads) — losing remarketing

Each of these costs you performance points, but they also generate revenue. Removing them to score 100 is like removing seats from a car to improve its 0-60 time.

Score Variability Makes It Meaningless

Lighthouse scores can fluctuate 5-15 points between runs due to:

• Network conditions
• CPU load on the test machine
• Third-party script timing
• Server response time variability

A page that scores 100 on one run might score 88 on the next. Obsessing over the exact number is chasing noise.

Lab Scores ≠ Real Performance

Lighthouse is a lab tool. A perfect lab score doesn't mean your users have a perfect experience. A page scoring 85 with great field data (CrUX) will outrank a page scoring 100 in lab with poor field data.

What Scores Actually Mean

Score Range	What It Means	Action
90-100	Good	Maintain and monitor
75-89	Decent	Optimize the biggest opportunities
50-74	Needs Work	Prioritize performance improvements
0-49	Critical	Performance is actively hurting you

Google's own documentation states that 90+ is considered "Good." There's no SEO benefit to scoring 100 vs. 92.

When Lower Scores Are Acceptable

E-Commerce Product Pages

A product page with high-quality images, reviews, and a dynamic pricing widget might score 70-80. That's fine — those features drive purchases. A fast but feature-stripped page converts worse.

SaaS Dashboards

Internal dashboards behind authentication don't need to score well on Lighthouse. They're not indexed by Google, and users accept slightly longer loads for powerful functionality.

Content-Rich Landing Pages

A landing page with an embedded video, interactive demos, and social proof elements might score 75. If it converts at 5% while a stripped-down 95-scoring version converts at 2%, the "slower" page wins.

The Right Approach: Performance Budgets

Instead of chasing a score, set performance budgets based on business-critical metrics:

Example Budget

Metric	Budget	Rationale
LCP	< 2.5s	Core Web Vital threshold
CLS	< 0.1	Core Web Vital threshold
INP	< 200ms	Core Web Vital threshold
Total JS size	< 300KB	Keep main thread responsive
Total image weight	< 1MB	Reasonable for content pages

If you stay within budget, the Lighthouse score will naturally be in the 85-95 range — which is exactly where you want to be.

The 90-Point Sweet Spot

Here's why 90 is the magic number:

1. Diminishing returns — Going from 50 to 90 has massive UX impact. Going from 90 to 100 has almost none.
2. Google's threshold — 90+ is "Good" in Google's framework.
3. Sustainable — You can maintain 90+ without sacrificing features.
4. Allows third-party tools — Analytics, chat, and tracking can coexist with a 90+ score.

What to Focus On Instead

Real User Metrics

Track what real users experience (field data) rather than what a synthetic test shows. CrUX data, RUM tools, and actual conversion rates matter more than a lab score.

Core Web Vitals Pass/Fail

The three Core Web Vitals (LCP, CLS, INP) are binary for SEO purposes — you either pass or fail. Focus on passing all three rather than maximizing a composite score.

Business Metrics

The ultimate performance metric is revenue. Track:

• Conversion rate by page speed segment
• Bounce rate correlated with LCP
• Revenue per session by performance bucket

Monitor Trends, Not Single Numbers

A Lighthouse score is a snapshot. What matters is the trend. Is your performance improving, stable, or degrading over time?

BadPageSpeed tracks your scores over time so you can spot trends and catch regressions — without obsessing over any single number.

Track performance trends →