The Chrome User Experience Report, known as CrUX, is a public dataset describing how real people using Google's Chrome browser experience websites: how quickly pages load, how fast they react to taps and clicks, and how much their layout jumps around. Google publishes it because laboratory tests run on fast machines in data centres say little about a mid-range phone on a congested network. CrUX supplies the field data behind Core Web Vitals, feeds Search Console and PageSpeed Insights, and, according to Chrome's developer documentation, is used by Google Search to inform its page experience ranking factor. As of today, it also measures advertising.

Who gets counted

Collection happens inside the browser, not through a page tag. According to Google's methodology documentation, a visit enters the dataset only when four conditions hold: usage statistic reporting is enabled, browsing history is synced to a Google account, no sync passphrase has been set, and the browser runs on a supported platform. Desktop Chrome on Windows, macOS, ChromeOS and Linux qualifies, as does Chrome on Android, including Custom Tabs and WebAPKs. Chrome on iOS, Android WebView and other Chromium browsers such as Microsoft Edge contribute nothing. Google does not publish what share of Chrome users meet the criteria.

Sites face two further tests. A page must be publicly discoverable under the indexability rules search engines apply, which rules out anything served with a status other than 200 or carrying a noindex directive. It must also be sufficiently popular. That threshold is secret; Google says only that it is identical for pages and origins and chosen to keep the distributions statistically reliable. Monitoring vendor DebugBear has estimated that URL-level data tends to appear at around 1,000 monthly views per device type, an observation rather than a published figure.

From page view to percentile

The primary unit is the origin, a scheme and host such as https://www.example.com, with page-level records beneath it. Query strings and fragments are stripped, so a URL tagged utm_medium=email merges with its untagged twin. Iframes are not reported separately; their layout shifts count against the embedding page, so a badly behaved ad frame can drag down a publisher's scores.

Metrics are stored as histograms whose bins sum to 1, alongside a 75th percentile value. The three Core Web Vitals are Largest Contentful Paint (LCP), which times the largest visible image or text block; Interaction to Next Paint (INP), which measures responsiveness; and Cumulative Layout Shift (CLS), a unitless stability score. A page counts as good when, at the 75th percentile, LCP is 2.5 seconds or less, INP 200 milliseconds or less and CLS 0.1 or less. Diagnostics include First Contentful Paint, Time to First Byte and round trip time. Dimensions include form factor (desktop, phone, tablet) and country. Google adds a small amount of random noise, which it calls fuzzing, so traffic volumes cannot be reverse-engineered, and any origin losing more than 20% of its traffic to ineligible dimension combinations is dropped entirely.

Every figure describes a 28-day rolling window, delivered on several clocks. The BigQuery dataset is refreshed on the second Tuesday of each month and named for the collection month, so release 202608 holds August 2026 data. The CrUX API updates daily at around 04:00 UTC. The CrUX History API refreshes weekly and, per current documentation, returns 40 weekly collection periods. PageSpeed Insights has drawn its field data from the CrUX API since December 3, 2020, and the Core Web Vitals report in Search Console is fed from the same pipeline.

Publishers, e-commerce teams and content management platforms are the main users, but the dataset carries a Creative Commons Attribution 4.0 licence, so anyone can query a competitor's origin.

Origins and evolution

Google introduced CrUX as a beta preview at Chrome Dev Summit on October 23, 2017, covering 10,000 origins. The November 2017 dataset lifted coverage past 1.2 million. A country dimension followed in January 2018, and First Input Delay (FID) arrived as an experimental metric in June 2018. CLS entered as an experiment in May 2019, and LCP launched that September.

The Web Vitals programme, introduced in May 2020, turned CrUX from a research dataset into the measurement layer for a ranking signal. The CrUX API launched in June 2020. Google's page experience update began reaching mobile search in mid-June 2021 and finished by the end of August, and the desktop version started rolling out on February 22, 2022, according to Google Search Central. The History API arrived in February 2023 with 25 weeks of data.

INP was added to CrUX as an experimental field with the April 2022 dataset and replaced FID as a Core Web Vital on March 12, 2024. Chrome then withdrew FID from the CrUX API, History API and BigQuery in September 2024. Effective connection type was retired as a dimension in the same period, replaced by a round trip time metric that widened coverage. Search Console removed its Page Experience report on November 18, 2024 while keeping the underlying Core Web Vitals report. CrUX Vis, a visualisation tool, launched experimentally in September 2024; the older Looker Studio dashboard was deprecated at the end of November 2025.

Coverage and scores have risen together. The June 2020 release counted about 7.5 million origins, 25.34% of them passing all three Core Web Vitals; August 2024 was the first in which a majority passed, at 50.1%.

Why marketers pay attention

Search is the obvious reason. Field data, not a laboratory score, decides whether a page qualifies as good, which is why a lazy-loading rollout that lifted a PageSpeed score from 65 to 92 still preceded a 20% traffic fall when LCP slipped from 1.8 to 4.2 seconds. How much weight the signal carries remains disputed. A survey of 131 SEOs found speed and Core Web Vitals among its most contested factors, and consultant Marie Haynes read Google's AI search documentation as treating middling scores as a minor driver outside e-commerce.

Revenue is the second. Google's AdSense team emailed publishers on March 24, 2026 citing an 18% ad revenue gain at German publisher Netzwelt after Core Web Vitals work, a figure relayed in Google's own communication. A study built on PageSpeed Insights found two-thirds of large US retail sites failing mobile speed tests before the 2025 holiday season.

Ad code appears in the release notes by name. After the Google Publisher Tag began yielding during out-of-viewport ad insertion in 2024, mobile INP pass rates for origins using it rose from 55.58% to 66.12%, according to the CrUX team. Work with consent management platforms including OneTrust was credited with INP gains in March 2024. Google also offers Publisher Ads Audits to trace the cost of ad code.

Blind spots and disputes

The sample is Chrome-only and opt-in, and its size is undisclosed. Chrome held 65.85% of global browser share in the first quarter of 2024 on Cloudflare's measurement, leaving the remainder unmeasured, along with Chrome on iPhone. The gap narrowed for site owners once Safari 26.2 shipped LCP and INP measurement in December 2025, allowing comparable figures from their own real user monitoring. Google's web.dev guidance concedes the two will diverge.

Single-page applications are poorly served. Route changes made by JavaScript look like new pages to a reader but are attributed to the first page view, a limitation Google blames on the underlying platform APIs. The hidden popularity threshold leaves low-traffic pages with no data, and the 28-day window delays the visible effect of any fix by about four weeks. Nor are movements always explained: the August 2026 release, published on September 8, 2026, called a continued INP regression a concern without a definitive cause.

The structural objection is concentration. The company that owns the dominant browser defines the metrics, collects the data and uses the result to rank the web. The Department of Justice prepared to seek Chrome's divestiture in November 2024, citing the signed-in user data the browser supplies.

Not the same as

Core Web Vitals are metrics; CrUX is one dataset reporting them. Lighthouse and other lab tools simulate a single load on controlled hardware, while CrUX aggregates real sessions. Real user monitoring relies on a script the site owner installs, covers every browser and exposes individual sessions; CrUX offers distributions only. Navboost click signals, stored in the Glue query logs a court ordered Google to share with rivals in September 2025, come from search results pages rather than page performance telemetry.

Recent developments

Google published documentation today for CrUX ad metrics, an experimental set of four measurements: Ad Count, Ad Density, Ad Weight: CPU and Ad Weight: Network. Chrome samples the viewport once per second to count visible ad frames and the share of the screen they occupy, and totals the processing and network load of ad scripts, reporting each at the 75th percentile. The CrUX API, History API and a DevTools panel expose the figures; BigQuery support is pending. Unlike Core Web Vitals, they carry no good or poor bands.

Eligibility introduces a supply-chain filter. Only origins whose ads.txt file names at least one authorised seller are included; sites with no file, or a placeholder record, are excluded. The rule leans on a format some argue is too coarse for modern selling. Density was already policed: the Better Ads Standards failed mobile pages above 30% ad density, Chrome filters ads on failing sites, and the browser has unloaded heavy ads since July 2020. What changes is visibility. A site's ad footprint, a variable that page-level yield metrics tend to reward, becomes a public and comparable number, measured by a browser owned by a company that also sells advertising.

Timeline

  • October 23, 2017: CrUX launches as a beta preview at Chrome Dev Summit with 10,000 origins
  • December 2017: The November 2017 dataset expands coverage past one million origins, with monthly releases promised
  • 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 launches in the dataset
  • May 2020: Google introduces Web Vitals
  • June 2020: CrUX API released; AMP Cache experiences attributed to publisher URLs from the June dataset
  • 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 on mobile, completing by end of August 2021
  • February 22, 2022: Page experience update begins rolling out on desktop
  • May 2022: Experimental INP appears in the April 2022 dataset
  • February 14, 2023: CrUX History API launches with 25 weeks of data
  • March 12, 2024: INP replaces FID as a Core Web Vital; navigation type breakdowns follow in April
  • September 2024: FID and effective connection type removed; CrUX Vis launches experimentally
  • September 10, 2024: August 2024 dataset shows a majority of origins (50.1%) passing Core Web Vitals
  • November 18, 2024: Search Console removes the Page Experience report
  • February 11, 2025: LCP subparts, LCP resource type and round trip time added
  • November 2025: CrUX Dashboard deprecated
  • December 2025: Safari 26.2 adds LCP and INP measurement APIs
  • September 8, 2026: August 2026 dataset covers 18,294,881 origins, 55.6% passing Core Web Vitals
  • September 15, 2026: CrUX ad metrics documentation published

Summary

Who: Google's Chrome team builds and publishes CrUX, drawing on Chrome users who have opted in to usage statistics and history sync. Publishers, e-commerce operators, SEO specialists, content management platforms and, increasingly, ad tech firms read it; Google Search uses it for page experience.

What: A public dataset of real-user performance measurements, aggregated by origin and page over 28-day windows and reported as histograms and 75th percentiles. It carries the Core Web Vitals (LCP, INP, CLS), diagnostic metrics and, since today, four experimental ad metrics.

When: Launched on October 23, 2017, expanded to more than a million origins that December, tied to ranking through the page experience update from June 2021, reshaped by the INP switch on March 12, 2024, and extended to advertising on September 15, 2026.

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

Why: Laboratory tests cannot show what real devices and networks deliver. CrUX provides that view at web scale, which gives it influence over search visibility and publisher revenue, and makes the single company that collects, defines and applies the data a recurring point of contention.