Chrome UX Report: Google User Speeds from A to Z

The Chrome UX Report (CrUX) is a public data set provided by Google, collecting and anonymously analyzing real user experience data from web browsing using Chrome.

Google's Chrome UX Report.

CrUX focuses on vital metrics such as loading speed, interactivity, and content stability, providing valuable insights into how websites perform under real-world conditions.

In addition to the basic Core Web Vitals (LCP, INP, CLS), CrUX now offers extended data that we've added to our Domain reports. This deeper insight helps you understand how your website behaves in real-world conditions. It includes user segmentation by device, the share of different page load methods, and details on what delays images in LCP.

CrUX vs. RUM and Synthetic#

CrUX isn't the only way to measure real user speeds. We've detailed the differences between measurement types, their strengths and weaknesses, and recommended combinations in the article synth vs. CrUX vs. RUM.

However, the Chrome UX Report is an authority for SEO and PPC. Google collects the data itself and uses it as a signal in search rankings and in Google Ads Quality Score.

How Are Data Measured in the Chrome UX Report?#

CrUX gathers data directly from the Chrome browser of users who haven't opted out of sharing web usage data. The domain and URL must have sufficient traffic for CrUX data to be displayed. Data are aggregated at the page and domain level and include additional information on connection speed, region, etc.

This makes CrUX data the only officially reported website speed source that impacts your ranking and ad cost. However, CrUX has limitations that must be considered:

• 28-day rolling average
  Technical changes impact CrUX values with a lag of several weeks. CrUX won't tell you "INP is better from today"; it shows a gradual curve trend.
• Device segmentation
  Data are collected separately for mobile, tablet, and desktop. The average user profile varies significantly between them.
• Visibility threshold
  A domain or specific URL must have sufficient traffic (estimated 10–15 thousand views per month) or data won't appear in CrUX.
• SPA vs. MPA
  For Single Page Applications, CLS and INP are measured cumulatively for the entire session, not just the initial load. CrUX values are often worse for SPAs compared to traditional MPAs, even if the "speed" feels similar. Details and measurement methods in Chrome are further discussed on web.dev's page about SPAs and Web Vitals.

For debugging specific issues, CrUX data isn't sufficient. Use synthetic or more advanced RUM measurement for all users.

Conversely, in PageSpeed.ONE, we use the Chrome UX Report as authoritative metrics for business decision-making (SEO effect, prioritization) and as a reference framework against which you compare synthetic and RUM results.

Why Is Data from the Chrome UX Report Important and Where Is It Used?#

CrUX is an essential resource for developers, SEO experts, and digital marketers because it offers insights from real user data. From our experience, the impact of speed has been most commonly observed in:

• SEO
  Google incorporates Core Web Vitals user experience metrics into website ranking in search results. Better performance can lead to better search rankings.
• PPC campaign evaluation
  Website speed affects bounce rates and the overall effectiveness of PPC campaigns. Speed can also influence the cost of ads in Google Ads.
• Speed measurement (UX)
  CrUX provides real data on page loading speeds, enabling developers to enhance user experience. By analyzing this data and optimizing, you can achieve faster websites, leading to higher user satisfaction. You can also compare your site with competitors.

What Metrics Does the Dataset Include?#

Google collects data separately for mobile, tablet, and desktop. CrUX contains information on the Core Web Vitals metrics, which include these basic metrics:

• Largest Contentful Paint (LCP)
• Interaction to Next Paint (INP)
• Cumulative Layout Shift (CLS)

Core Web Vitals Metrics: the most fundamental of all web speed metrics.

Further values available in the dataset are discussed below.

Origin vs. URL: Data Granularity#

The Chrome UX Report provides data for two page levels. The difference is crucial for proper interpretation:

• Origin (e.g., https://www.example.com) returns aggregated data across the entire source, including all subpaths and localizations (/en-us, /fr, /product/…). Origin gives a cross-section of the entire domain.
• URL (e.g., https://www.example.com/en-us) returns data only for this specific URL. It does not include subpages (/en-us/product/abc).

For grouping URLs by pattern, use Google Search Console, which operates with its own URL group segmentation. A guide on how to read its speed reports is available at Google Search Console and Core Web Vitals.

In practice, you may encounter situations where CrUX data is missing, typically due to three reasons:

1. Insufficient traffic
   Without ~10–15 thousand PV per month, data won't appear in CrUX, either for the entire domain or for specific URLs.
2. noindex on the homepage
   If the homepage is marked noindex, CrUX won't report origin data.
3. Redirect from the root page
   If the root automatically redirects to a sub-page (e.g., domain.com to domain.com/welcome) and the original root has noindex, CrUX follows the original noindex and may not follow the redirect. Result: origin data disappears, even if the /welcome itself is perfectly fine.

Advanced Metrics and Navigation Types#

CrUX today offers more than just the three Core Web Vitals metrics. Among the advanced data, these are most utilized:

• RTT (Roundtrip Time) separates the backend impact from the network impact. If you have high RTT but reasonable TTFB, the user's connection to the server is slow, and considering a CDN makes sense.
• LCP Image Subparts reveal where time is lost in the LCP image. Breaking down into server delay, download time, and rendering helps identify whether a better backend, smaller image, or fetchpriority will help.
• LCP Resource Type distinguishes whether the LCP is an image or text/webfont, which determines where to focus optimization efforts. Either preload the image or preload the critical font.
• Navigation Types show the share of different navigation types. Increasing the share of BFcache and pre-rendering through Speculation Rules shifts LCP and INP values towards zero, as "instant" navigations are counted as fast in CrUX.

Tracking the share of different navigation types helps us see the proportion of fast or instant navigations.

In PageSpeed.ONE practice, we implemented Speculation Rules with Innogy, where the increasing share of instant navigations significantly improves values for the entire domain. The recipe for keeping the INP metric under control can be found in INP Optimization.

Data Delays in CrUX#

Data in the Chrome UX Report are not real-time. They enter CrUX with a delay of up to a month from the end of the reporting period, depending on the dataset:

• Monthly dataset (CrUX BigQuery) is published on a fixed day (e.g., the second Tuesday of the month) and includes data for the previous calendar month.
• CrUX Daily is more current. Today, you'll see data for yesterday. However, it uses a 28-day rolling window.

Histogram of Metric Value Distribution#

The 75th percentile value of a metric (e.g., 2.1 s for LCP) and the three-color categories (portion of users for whom the metric is satisfactory, needs improvement, unsatisfactory) are merely summaries of the overall speed picture for users.

The histogram goes a level further. It shows what portion of users (or more precisely, page views) experienced a specific metric value in the last month. It thus indicates how many users experienced the metric at a given value.

It's an advanced view. It will make sense if you know how to read percentiles, distribution shapes, and think about speed optimization at the user segment level, not just at a single number level. For daily monitoring, the state of metrics and their development over time suffices; consider the histogram as a diagnostic detail.

Histogram in the Domain Report PageSpeed.ONE.

Where Can You Find the CrUX Histogram?#

In PageSpeed.ONE, you'll find the histogram in the Domain Report. The same type of view (distribution of values from CrUX) is also available in other tools like Treo or in the source dataset for the Chrome UX Report itself.

Why Monitor the Histogram?#

The 75th percentile is a single number. The histogram shows whether the bulk of users is just below the threshold, whether a green 75th percentile hides a large orange or red mass, or whether the problem is a thin "tail" of extremely slow views. Without knowing the distribution shape, you might optimize the wrong segment.

How to Read the Shape of a CrUX Histogram?#

The histogram shows the last month's CrUX data without a period switch. It's not about comparing "before and after" in this graph – that's what daily and monthly developments are for. From a single snapshot, you can discern:

• Close to the threshold vs. deep in red – most "bad" values just beyond the limit often mean a minor tweak suffices; a mass far into the red zone suggests website structure or template issues.
• Multiple peaks – typically different segments (cache vs. no cache, various templates, geographical differences). A classic double distribution for TTFB is described in Backend Optimization.
• Thin right tail – a few catastrophic views vs. a shift of the entire mass to the right (problem across the site).
• Metric by metric – for LCP, focus on any second peak; for INP, a long tail of heavy interactions; for CLS, a peak near zero and a thin tail vs. an entire distribution shift.
• Domain comparison – same 75th percentile and different shape = different optimization priorities (majority slow vs. majority fine and a few disasters).

The Relationship of the Histogram to the PageSpeed.ONE Score#

The PageSpeed.ONE Score (SPS) already considers the entire distribution of users, not just the 75th percentile. The histogram provides a visual drill-down for individual metrics: you see the shape from which SPS (and Google) derive.

Where Can You Find Data from the Chrome UX Report?#

Numerous tools work with CrUX data. They differ in currency, metric support, and clarity. In practice, we recommend starting with our web speed test tool (Insights). You'll get a clear UI, PageSpeed.ONE Score (SPS) as a single summary number, CrUX data development over time, and the integration of synth and CrUX data in one dashboard:

Other options for obtaining CrUX data include:

• CrUX Vis is an experimental Google tool for CrUX visualization. It's more suited for specialists who want flexibility at the expense of clarity.
• Google's PageSpeed Insights shows data but doesn't provide time development. It's fine for one-time verification, not for monitoring progress.
• Google BigQuery provides raw data for advanced analysis using SQL. There is also a CrUX History API, which returns weekly data and is faster than ad-hoc BigQuery queries.

How Do We Use CrUX Data in PageSpeed.ONE Monitoring?#

CrUX data reports from PageSpeed.ONE's PLUS monitoring display the status of metrics on the dashboard and the development of values on domains:

Speed development according to CrUX data in speed monitoring from PageSpeed.ONE.

Monitoring CrUX data is very important for all websites that have these data. In our Domain Report, you'll find among other things:

• Core Web Vitals metrics (LCP, INP, CLS).
• Additional speed metrics (FCP, TTFB, RTT).
• Histogram of metric value distribution – a detailed look at the density of values over the past month.
• User segmentation by device type (mobile, tablet, desktop).
• Segmentation of page load methods (navigate, reload, bfcache, prerender…).
• Source type for the LCP metric (image vs. text).
• Reasons for delays in images that are the LCP element,
• Development of these data by days and months.

These pieces of information together form a picture of how your users truly experience website speed and where there is room for improvement. If you prefer to track a single summary number instead of individual metrics, we recommend PageSpeed.ONE Score (SPS), which is derived from CrUX data and considers the entire user distribution, not just the 75th percentile.