How Lens scores work

Here’s what each score means — and what it doesn’t. Weights reflect business impact. Limitations are named so you can trust the report without mistaking a snapshot for continuous monitoring.

Overall score

A weighted average of four dimension scores. Weights reflect estimated business impact — not equal splits.

Confidence system

Scores are not equally reliable at all sample sizes. With fewer pages analyzed, scores are pulled toward neutral (50) and labeled directional — not definitive. Three or more pages analyzed produces a high-confidence score.

Benchmark data

Every score is contextualized against industry reference percentiles built from our own Lighthouse audits of representative sites in each industry, refreshed regularly. Where available, speed scores also incorporate real-user field data from the Chrome UX Report — measurements from real Chrome users, not lab simulations. These are not scores from named companies — they reflect population-level real-world data.

Known limitations

• Server-side tagging: Lens detects client-side tag signatures only. GTM server-side or Segment server-to-server implementations will not be detected and will lower your Signal Coverage score even if analytics are fully instrumented.
• Single-session performance: CWV scores reflect your browser, network, and moment in time. Use Google Search Console field data for production decisions.
• Structural vs. behavioral: Conversion Architecture scores page-type presence in your sitemap — not whether users actually convert.