CrUX, short for the Chrome User Experience Report, is a public dataset of performance measurements taken from real people browsing with Google Chrome. It exists because laboratory tests run on a developer's machine cannot show what happens on a mid-range Android phone over a congested mobile network. Chrome records timings as pages load and respond, Google aggregates them by website and by page, and the results are published monthly and daily. The dataset is the source of field data in PageSpeed Insights and Search Console, and, according to Google's documentation, it informs the page experience ranking factor in Google Search.

From browser to dataset

Collection happens inside the browser, not through a tag on the page. According to Chrome's methodology documentation, a user's visits count only if four conditions hold: usage statistic reporting is enabled, browser history is synced, no sync passphrase is set, and the platform is supported. Desktop Chrome on Windows, macOS, ChromeOS and Linux qualifies, as does Chrome on Android, including Custom Tabs. Chrome on iOS, Android apps using WebView and other Chromium browsers such as Microsoft Edge contribute nothing. Google does not publish the share of users meeting these criteria.

Websites face two further tests. A page must be publicly discoverable, judged by the same indexability rules search engines apply: a non-200 status after redirects, an X-Robots-Tag noindex header or a robots noindex meta tag each disqualify it. It must also be sufficiently popular, a visitor threshold Google declines to disclose. The same minimum applies to pages and to origins, the protocol-plus-hostname unit that represents a whole site.

Processing then shapes what survives. Query strings and fragments are stripped, so a URL tagged with campaign parameters is merged with its clean version. Iframes are not reported separately; their behaviour counts against the top-level page, which means a third-party ad slot that shifts the layout lowers the publisher's own score. An origin losing more than 20% of its traffic to ineligible dimension combinations is dropped entirely, and a small amount of random noise is added so total traffic cannot be reverse-engineered.

What gets measured

The core outputs are the three Core Web Vitals. Largest Contentful Paint (LCP) times the rendering of the main content, with 2.5 seconds as the "good" boundary. Interaction to Next Paint (INP) measures responsiveness, good below 200 milliseconds. Cumulative Layout Shift (CLS) scores visual instability, good at 0.1 or less. Supporting metrics include First Contentful Paint, Time to First Byte, navigation types and round trip time.

Values are stored as histograms, and the headline figure is the 75th percentile, or p75. A page passes a metric only when three quarters of recorded experiences meet the threshold, a convention written into Joost de Valk's 128-rule website specification in May 2026. Data can be segmented by form factor (phone, desktop, tablet), by country in BigQuery, and by a popularity rank expressed in half-steps from the top 1,000 origins upwards.

Four ways in

Access comes through several surfaces fed by the same pipeline. PageSpeed Insights has sourced its field data from the CrUX API since December 3, 2020, and Google told developers in March 2024 to use it, or a real user monitoring vendor, to check eligibility for the INP transition. The CrUX API returns a rolling 28-day window, updated daily with a two-day lag, and is capped at 150 queries per minute per Google Cloud project without charge. The CrUX History API returns weekly snapshots, 25 by default and up to 40, roughly ten months. BigQuery holds monthly origin-level tables, released on the second Tuesday of each month. Search Console groups URLs by CrUX status, and Chrome DevTools now overlays field data on local traces.

Origins and evolution

Google introduced a beta at the Chrome Dev Summit in October 2017 covering 10,000 origins. The next release expanded coverage to 1,237,407. A country dimension followed in January 2018, First Input Delay (FID) arrived as an experimental metric in June 2018, CLS in May 2019 and LCP in September 2019.

Web Vitals were announced in May 2020, and Google said that month it would combine them with mobile-friendliness, HTTPS and interstitial rules into a ranking signal. After a delay from May, the page experience update began rolling out on June 15, 2021 and finished by August 31, 2021. From that point the dataset carried commercial weight for anyone dependent on organic traffic.

The metric set kept moving. An experimental INP entered in May 2022, releases stopped requiring connection type and form factor that year and origin coverage rose 28.2%, and the History API launched in February 2023. INP replaced FID as a Core Web Vital on March 12, 2024; FID and the effective connection type dimension were removed from CrUX tooling in September 2024, with round trip time taking the latter's place. Google launched CrUX Vis the same month and retired the older Looker Studio dashboard at the end of November 2025.

Why marketers read it

Pass rates have climbed but remain unremarkable. In the August 2026 release, published September 8, 2026, 55.6% of 18,294,881 origins had good Core Web Vitals, against roughly 22% in February 2021. Sector studies built on the same data are harsher: a 20i review of more than 350 US retailers found 76% of retail and wholesale sites failing in October 2025.

The revenue case is usually argued through conversion. Shopify's analysis, summarised in PPC Land's latency explainer, linked each additional 100 milliseconds of main-content load to roughly 3.5% lower conversion. Publishers hear the argument from their monetisation partner too: an AdSense email on March 24, 2026 pressed bfcache and speculation rules, citing a 27% page view gain at Netzwelt.

Ad technology shows up in the data directly. According to CrUX release notes, mobile INP pass rates for origins running Google Publisher Tag rose from 55.58% to 66.12% after an August 2024 change that yielded before inserting out-of-viewport ads. Earlier that year, consent management platforms including OneTrust, Complianz and Axeptio were credited with INP gains after changing how often their scripts yield.

Limitations and disputes

The sample is Chrome-only and self-selected, which excludes most iPhone users. For years that asymmetry meant Core Web Vitals could only be measured reliably in one browser; Safari 26.2, Firefox 146 and Chrome 143 closed the gap in December 2025, but CrUX itself still draws on Chrome alone. The undisclosed popularity threshold leaves smaller sites with no data at all.

Ranking weight is contested. Google's Martin Splitt described Core Web Vitals in August 2025 as a tiny factor in most cases, while practitioners report the opposite experience: one agency attributed a 20% traffic loss to lazy loading that pushed LCP from 1.8 to 4.2 seconds even as the lab score improved. Some performance vendors also claim CrUX feeds Google Ads Quality Score; that link rests on vendor assertion rather than Google documentation reviewed for this article.

Governance is the larger question. Google owns the browser that collects the data, defines the metrics and uses them in ranking. The Justice Department sought a Chrome sale, which Judge Amit Mehta rejected on September 2, 2025. Google also removed the consolidated Page Experience report from Search Console on November 18, 2024, shifting interpretation back to individual reports.

Technical quirks persist. Single-page applications attribute every soft navigation to the first page view, and Google's 28-day window means fixes take weeks to register.

Not the same as

Core Web Vitals are the metrics; CrUX is one dataset that reports them. Any real user monitoring tool can measure the same values across all browsers.

PageSpeed Insights and Lighthouse produce lab scores from a simulated load. PageSpeed Insights displays CrUX field data alongside, and the two often disagree.

Real user monitoring (RUM) uses a script the site owner installs, capturing every browser and every visit, with no public benchmark.

Ad Experience Report and Heavy Ad Intervention are Chrome enforcement tools. The first applies the Coalition for Better Ads standards and can filter all ads on a failing site; the second, live since July 14, 2020, unloads ads exceeding 4MB of network data or heavy CPU thresholds. CrUX measures without enforcing.

Recent developments

Chrome published documentation today for CrUX ad metrics, an experimental set covering Ad Count, Ad Density, Ad Weight: CPU and Ad Weight: Network, available at origin and page level through the CrUX API, the History API and a DevTools Ads panel, with BigQuery support in preparation. According to Chrome's overview, the aim is to help advertisers and sites value ad opportunities. The metrics use the same p75 convention but carry no good or poor labels and sit outside Core Web Vitals.

Measurement relies on viewport sampling once per second and on Chrome's ad detection, which matches requests against a condensed version of EasyList, the open-source filter list used by ad blockers, and flags resources requested by ad scripts. Only origins declaring at least one authorised seller in ads.txt are included, a gate that ties browser telemetry to a 2017 file format now facing a replacement proposal in adagents.json. Ad Density echoes a threshold in PPC Land's ad load explainer: the Coalition's research failed mobile pages above 30%.

Meanwhile, the August 2026 release noted continued INP regression, which Google said it could not yet explain.

Timeline

  • October 2017 - Chrome User Experience Report beta launches at Chrome Dev Summit with 10,000 origins
  • November 2017 - Dataset expands to more than 1.2 million origins
  • January 2018 - Country dimension added
  • June 2018 - First Input Delay added as an experimental metric
  • May 2019 - Cumulative Layout Shift added as an experimental metric
  • September 2019 - Largest Contentful Paint launched
  • May 2020 - Google announces Web Vitals and a forthcoming page experience ranking signal
  • December 3, 2020 - PageSpeed Insights begins sourcing field data from the CrUX API
  • February 2021 - Coarse popularity rank added to BigQuery
  • June 15, 2021 - Page experience ranking update begins rolling out, completed by August 31, 2021
  • May 2022 - Experimental Interaction to Next Paint added; form factor and connection type no longer required
  • February 2023 - CrUX History API launched
  • March 12, 2024 - INP replaces FID as a Core Web Vital
  • April 2024 - Navigation type breakdowns added
  • September 2024 - FID and effective connection type removed; round trip time introduced; CrUX Vis launched
  • November 18, 2024 - Page Experience report removed from Search Console
  • February 11, 2025 - LCP subparts and resource types added to the CrUX API
  • November 2025 - CrUX Dashboard deprecated
  • September 8, 2026 - August 2026 release covers 18,294,881 origins, 55.6% with good Core Web Vitals
  • September 15, 2026 - Chrome publishes documentation for experimental CrUX ad metrics

Summary

Who: Google's Chrome team operates CrUX, with contributions from Chrome users who sync history and share usage statistics. Site owners, developers, SEO specialists, publishers and, with the new ad metrics, advertisers read the data through Google and third-party tools.

What: A public dataset of real-user performance measurements, aggregated by origin and page over rolling 28-day windows and reported as histograms and 75th-percentile values for Core Web Vitals and supporting metrics, now joined by experimental ad count, density and weight metrics.

When: Launched in beta in October 2017, tied to Search ranking from June 2021, reshaped by the INP switch in March 2024, and extended to ad experiences on September 15, 2026.

Where: Collected from Chrome on desktop and Android worldwide, excluding iOS and WebView, and published through PageSpeed Insights, Search Console, the CrUX API and History API, BigQuery, CrUX Vis and Chrome DevTools.

Why: Laboratory tests cannot reproduce real devices and networks, so Google built a field benchmark that anyone can query. Its role in ranking, its Chrome-only sample and Google's control of both the browser and the metrics make it influential and contested in equal measure.