# Your Lighthouse Score Is Green But Real Users Are Suffering Lighthouse tests a single cold load with throttled network on a simulated device. CrUX aggregates 28 days of real-user sessions. When they disagree — lab green, field red — your synthetic test has a blind spot. The tool queries Google's PSI API for both, computes the delta per metric, and emits the diagnosis prompt. Author: J.A. Watte Published: May 11, 2026 Source: https://jwatte.com/blog/blog-tool-cwv-field-vs-lab-gap/ --- _Part of the [performance audit tool stack](/blog/blog-new-performance-audit-tools-2026/). See the pillar post for the full catalog of sibling audits and where this one fits in the lineup._ Lighthouse measures one controlled load. CrUX measures 28 days of reality across thousands of real sessions. They disagree more often than teams realize. Typical lab blind spots: - **INP.** Lab measures initial-load interaction. Field captures mid-session INP after the user has been clicking around for minutes. Third-party scripts that lazy-load on scroll degrade field INP while passing lab INP. - **CLS.** Lab measures above-the-fold layout shift on first paint. Field accumulates CLS over the whole session — infinite-scroll injections, delayed-font-swap shifts, ad refreshes all contribute. - **LCP.** Lab uses one geography and one device profile. Field mixes fast and slow, mobile and desktop. A site with a p25 user base on 3G will have field LCP far worse than lab. [CWV Field vs Lab Gap Audit](/tools/cwv-field-vs-lab-gap/) queries Google's PageSpeed Insights API for both data sources in one call, computes the delta per metric, and flags the lab blind spots. ## Using the tool Enter a URL and a strategy (mobile or desktop). The tool hits PSI's public endpoint (no key required for light use; get a free API key for higher rate limits). Returns: - **Per-metric lab value** (from Lighthouse audit). - **Per-metric field p75** (from CrUX 28-day aggregate for the URL, or origin-level if URL has insufficient traffic). - **Gap classification**: aligned, lab blind spot (field worse), or field ahead (lab worse). The lab-blind-spot metrics are the ones to prioritize — real users are the ones who matter for both conversion and ranking. ## What to do about each blind-spot pattern - **INP lab < 200ms but field > 500ms.** Deploy Real-User Monitoring (RUM) via the web-vitals.js library. Attribution-mode RUM surfaces which event handlers are slow in production. Fix third-party scripts lazy-loading on scroll by deferring or unloading them. - **CLS lab 0.05 but field 0.25.** Your session has shifts lab doesn't catch. Common cause: ad refresh inserting without aspect-ratio, infinite-scroll element insertion, font-swap flash. - **LCP lab 1.8s but field 4.2s.** Your real user base has slower network / device than lab simulates. Ship priority hints on the LCP image, preload the LCP image, optimize the critical-path CSS. ## Why this matters for ranking Google uses CrUX (field) data for ranking, not Lighthouse (lab) data. Your lab score could be 100/100 and your ranking would still suffer if real users are hitting timeouts. This audit is the diagnostic pass that connects lab optimization to real-user impact. ## Related reading - [CWV Audit](/tools/cwv-audit/) — local measurement - [CWV Fix Generator](/tools/cwv-fix-generator/) — emits remediation patches - [INP Attribution](/tools/inp-attribution/) — per-handler attribution ## Fact-check notes and sources - PSI API v5: [developers.google.com/speed/docs/insights/v5/get-started](https://developers.google.com/speed/docs/insights/v5/get-started) - Chrome UX Report (CrUX): [developer.chrome.com/docs/crux](https://developer.chrome.com/docs/crux) - Web Vitals library: [github.com/GoogleChrome/web-vitals](https://github.com/GoogleChrome/web-vitals) --- *The $100 Network covers CWV as a site-network concern — where one template change lifts 50 sites at once. The field-vs-lab audit is how you prove real users are getting the lift.* --- Canonical HTML: https://jwatte.com/blog/blog-tool-cwv-field-vs-lab-gap/ RSS: https://jwatte.com/feed.xml JSON Feed: https://jwatte.com/feed.json Hero image: https://jwatte.com/images/blog-tool-cwv-field-vs-lab-gap.webp