---
title: Your Lighthouse Score Is a Lab Test. The CrUX API Shows What Real Visitors Feel.
description: Google ranks you on field data from real Chrome users, not your last Lighthouse run. Here is how to pull CrUX data with one free API call and monitor your templates and competitors weekly.
author: LETSGROW Dev Team
date: 2026-10-29
category: Analytics
tags: ["CrUX API", "Core Web Vitals", "Web Performance", "Analytics", "SEO"]
url: "https://letsgrow.dev/blog/crux-api-field-data-monitoring"
---
# Your Lighthouse Score Is a Lab Test. The CrUX API Shows What Real Visitors Feel.

Your Lighthouse score is a lab test, and Google does not rank you on lab tests. Google judges your site on field data collected from real Chrome users, and almost every marketing team is monitoring the wrong number. The fix is a free API, a small script, and about an hour of setup.

## Lab Data Tells You What Could Happen. Field Data Tells You What Did.

Lighthouse and PageSpeed Insights lab runs load your page once, on a simulated device, on a simulated network. That is useful for debugging. It is useless for answering the question that matters: how do your actual visitors experience the site?

The Chrome User Experience Report, or CrUX, answers that question. It aggregates real-user measurements of LCP, INP, CLS, FCP, and TTFB from opted-in Chrome users over a rolling 28-day window, and Google's page experience signals draw on this dataset, not on your last Lighthouse run. Marketing teams that report a green lab score while their field INP sits at 300ms are reporting fiction.

The practical gap is large. A lab run cannot see the consent banner that loads late for visitors in Europe, the personalization script that only fires for logged-in users, or the third-party chat widget that hammers low-end Android phones. Field data sees all of it, because it comes from the people who actually hit those paths.

::stat-block
title: "Lab Data vs Field Data"
stats:
  - value: "28 days"
    label: "rolling window of real Chrome user data behind every CrUX number"
  - value: "p75"
    label: "the percentile Google uses to judge each Core Web Vital, so your slowest quarter of visits decides your result"
  - value: "Free"
    label: "cost of the CrUX API, which needs only a Google Cloud API key"
::

## Pull It With One Request

The CrUX API is a single POST endpoint. You send an origin or a specific URL, an optional form factor, and a list of metrics. You get back histograms and the p75 value for each.

```js
const res = await fetch(
  `https://chromeuxreport.googleapis.com/v1/records:queryRecord?key=${KEY}`,
  {
    method: 'POST',
    headers: { 'Content-Type': 'application/json' },
    body: JSON.stringify({
      origin: 'https://www.example.com',
      formFactor: 'PHONE',
      metrics: ['largest_contentful_paint', 'interaction_to_next_paint', 'cumulative_layout_shift'],
    }),
  }
);
const { record } = await res.json();
console.log(record.metrics.interaction_to_next_paint.percentiles.p75);
```

Always set formFactor. Phone and desktop distributions differ sharply, and a blended number hides the mobile problem that usually drives the ranking risk. Run the query for PHONE and DESKTOP separately and store both.

Two limits to know up front. CrUX only returns data for origins and URLs with enough traffic, so low-traffic pages return a 404, and your script must treat that as "no data" instead of an error. And the standard endpoint is a snapshot, so for trends you want the history endpoint, `records:queryHistoryRecord`, which returns weekly collection periods you can chart without storing anything yourself.

## Monitor Templates and Competitors, Not Just the Homepage

The homepage is the page you least need to monitor. It gets the most engineering attention already. The pages that quietly lose rankings are the templates nobody owns: the blog post layout, the pricing page, the landing page builder output.

Build a watchlist with three groups:

1. Your origin, split by phone and desktop
2. Five to ten high-traffic URLs representing each template type
3. Two or three competitor origins

The competitor group is the part most teams skip, and it is the most persuasive in a planning meeting. CrUX data is public for any origin with enough traffic. A chart showing your phone LCP against the three sites you compete with for the same keywords turns "performance work" from an engineering favor into a ranking argument.

Schedule the script weekly, write results to a BigQuery table or even a spreadsheet, and alert when any p75 crosses a threshold: LCP above 2.5 seconds, INP above 200 milliseconds, CLS above 0.1. Those are Google's published "good" boundaries.

::checklist
title: CrUX Monitoring Setup
items:
  - Create a Google Cloud API key restricted to the Chrome UX Report API
  - Query PHONE and DESKTOP separately for your origin, never the blended default
  - Add one representative URL per template type, not just the homepage
  - Add two or three competitor origins to the same weekly job
  - Treat a 404 response as "insufficient traffic", not as a failure
  - Store p75 values with a timestamp so you can chart week over week
  - Alert on LCP over 2.5s, INP over 200ms, and CLS over 0.1
  - Annotate the chart with deploy dates and tag changes
::

## Connect Regressions to Releases

A chart of p75 values is a thermometer. It becomes a diagnostic tool when you annotate it with what changed. Log every deploy and every new marketing tag in the same table. When INP jumps from 180ms to 260ms the week after a new tag manager container went live, you have your culprit without a single profiling session.

Keep the lab tools, but demote them. Use Lighthouse to find why a page is slow, and use CrUX to decide whether it matters. If field data is green, a red lab score is a debugging curiosity. If field data is red and the lab score is green, your lab setup is not reproducing what your users are living through, and you should instrument real-user monitoring with the web-vitals library to find the gap.

## The Takeaway

Stop reporting Lighthouse scores to stakeholders. Replace that slide with a weekly CrUX chart: your phone p75 values for the three Core Web Vitals, your key templates, and your competitors on the same axis. It costs nothing, it measures what Google measures, and it will tell you about a regression a week after it ships instead of a quarter later when traffic has already dipped.
