Release notes

The CrUX dataset on BigQuery is updated on the second Tuesday of every month. Each release is numbered according to the year and calendar month of the data collection period, for example 201912 corresponds to the UX data collected during December 2019 and would be released on the second Tuesday of January 2020 after the data collection period has ended.

In the following list, we've curated some release notes for each monthly dataset. Subscribe to our CrUX Announce mailing list or follow @ChromeUXReport on Twitter for release Announcements.

202407

Announcement

Publication date

  • August 13, 2024

What's new

  • This month we saw another good increase in Core Web Vitals pass rates across all metrics.
  • There are a number of changes being worked on by the Chrome team (particularly for LCP and INP) which no doubt helped with these improvements, including the two changes to INP mentioned last month which should be fully rolled out now. We encourage those interested to check the Web Vitals Changelogs for more technical detail, and will continue to highlight particularly noteworthy changes in these monthly announcements.
  • This month we did a small tweak to our CrUX API documentation, and brought back the CrUX API Explorer that we used to have on our documentation in the past, but which we had to remove for technical reasons. This allows readers of the docs to quickly check API responses. By viewing this explorer in full screen, you even get cURL, HTTP, and JavaScript snippets.
  • This is also your final reminder that FID is deprecated and will be removed from Chrome tools in September 2024. Make sure you update any CrUX API applications to switch over to INP by then.

Notable stats

  • 18,176,653 origins (↓ 1.4%)
  • 64.1% of origins (↑ 1.2%) had good LCP
  • 78.0% of origins (↑ 0.4%) had good CLS
  • 84.4% of origins (↑ 0.4%) had good INP
  • 49.2% of origins (↑ 1.6%) have good Core Web Vitals

202406

Announcement

Publication date

  • July 9, 2024

What's new

  • This month we saw a good increase in Core Web Vitals pass rates across all metrics—a reverse of the slight regression we saw last month. If you've followed us carefully you'll be aware that some of the shifts may be seasonal, for example, in July 2023 we also saw a decrease in the number of origins, and an increase in pass rates for the Core Web Vitals.
  • The Chrome team has been continuing work on improving efficiencies in Chrome's handling of the Core Web Vitals metrics and recently launched some changes to INP, most notably to better handle use of the basic modal dialogs (alert, confirm, print). While technically these are synchronous and block the main thread—and so are not recommended if there are alternatives—they do present user feedback for an interaction. They were previously not counted as presentation feedback for INP, which could result in very high INP values for sites that did use these. From Chrome 127 the presentation of the modal will mark the end measurement time for INP and so should lead to improved INP times for those sites.
  • As previously mentioned, we're planning to remove the Effective Connection Type (ECT) dimension from CrUX in September 2024. With this release, we're including a preview of the new round_trip_time (RTT) metric in the CrUX API, which measures network latency. See our announcement for details.
  • And finally, as a reminder, FID is deprecated and will be removed from Chrome tools in September 2024. Make sure you update any CrUX API applications to switch over to INP by then.

Notable stats

  • 18,442,199 origins (↓ 1.2%)
  • 63.4% of origins (↑ 2.0%) had good LCP
  • 77.8% of origins (↑ 0.5%) had good CLS
  • 84.1% of origins (↑ 1.1%) had good INP
  • 48.4% of origins (↑ 2.6%) have good Core Web Vitals

202405

Announcement

Publication date

  • June 11, 2024

What's new

  • We have seen a slight regression in LCP this month, which has affected the overall Core Web Vitals pass rate. This seems to have particularly affected mobile scores (desktop LCP scores are actually slightly up). We've been unable to find a specific reason for this change, but this does come on the back of an impressive run of form with this only being the second negative month in the last year. So at this stage we're not concerned with this slight drop this month.
  • We want to give you advance notice for a change we are considering to make in three months (for the September 10th release). We are looking to drop the Effective Connection Type (ECT) dimension since it is proving less useful as more and more internet traffic has moved to the wide-ranging 4G bucket. Having this as a dimension leads to other issues including a reduced coverage for origins and data. However, we do think measuring latencies on the web is important and are investigating replacement options, but as a metric rather than a dimension (so the other metrics will no longer be able to be queried by ECT). We would like to receive feedback from the current users of this data to understand how impactful this would be and if we need to consider more time for the migration. Let us know by responding to this email if you would be impacted by this change.

Notable stats

  • 18,673,241 origins (↓ 0.2%)
  • 62.1% of origins (↓ 0.5%) had good LCP
  • 77.4% of origins (~ 0.0%) had good CLS
  • 83.2% of origins (↑ 0.5%) had good INP
  • 47.1% of origins (↓ 0.3%) have good Core Web Vitals

202404

Announcement

Publication date

  • May 14, 2024

What's new

  • No significant updates in this release.

Notable stats

  • 18,703,230 origins (↑ 0.2%)
  • 62.4% of origins (↑ 0.3%) had good LCP
  • 77.5% of origins (↑ 0.4%) had good CLS
  • 82.8% of origins (↑ 0.8%) had good INP
  • 47.3% of origins (↑ 1.1%) have good Core Web Vitals

202403

Announcement

Publication date

  • April 09, 2024

What's new

  • All metrics have seen a good improvement this month, more than making up for last month's small declines. This is particularly apparent on mobile.
  • We're aware of a two changes which resulted in INP improvements for a large number of sites this month:
    • The Chrome Android team have made improvements to cookie caching, which saw noticeable improvements to INP and LCP. We think this was a large factor in the improvements we have seen this month.
    • The team at Google have been working with a number of Consent Management Platforms (including OneTrust, Complianz and Axeptio), to improve INP by yielding more often—particularly when cookies are accepted. This has resulted in much improved INP for sites using these platforms.
  • This month we are pleased to announce that we have added navigation type breakdowns to CrUX. By exposing the navigation type breakdown, we hope to encourage site owners to be more aware of the navigation types used on their sites, and look to encourage some of the faster types by looking at caching setup, bfcache blockers, and prerendering. For more details are available in the navigation types announcement post.
  • We have also changed the CrUX API and CrUX History API to always round floating point numbers to 4 decimal places.

Notable stats

  • 18,669,191 origins (↓ 0.3%)
  • 62.2% of origins (↑ 1.5%) had good LCP
  • 77.1% of origins (↑ 0.6%) had good CLS
  • 82.1% of origins (↑ 1.8%) had good INP
  • 46.8% of origins (↑ 2.6%) have good Core Web Vitals—a reminder this is now measured based on INP and not FID.

202402

Announcement

Publication date

  • March 12, 2024

What's new

  • The big news this month is that INP is replacing FID from today as a Core Web Vital metric. This will be the last month that we report the FID numbers in these announcement reports. Tooling such as PageSpeed Insights and the CWVTech Report have been switched over to use INP when measuring Core Web Vitals pass rates using CrUX from today. FID should be considered deprecated now and will be removed from the Chrome tools, including CrUX, on 9th September 2024.
  • We have regressed in most of the metrics this month by small amounts. We have not been able to ascertain any particular reason for this, and overall the metrics are still up from 2 months ago and previous, so we are treating these as more a slight "return to norm" of last month's impressive improvements rather than anything to be concerned about.

Notable stats

  • 18,583,879 origins (↑ 0.8%)
  • 48.8% of origins (↓ 0.4%) have good Core Web Vitals when using FID
  • 45.6% of origins (↓ 0.4%) have good Core Web Vitals when using INP in place of FID as will be the case from today onwards.

202401

Announcement

Publication date

  • February 13, 2024

What's new

  • The seasonal drop in origins from December has fully recovered.

Notable stats

202312

Announcement

Publication date

  • January 09, 2024

What's new

  • Most metrics improved this month with the exception of a very slight decrease in FID and INP. Increases and decreases in pass rates were most likely impacted by the larger than usual swing in origins this month rather than any changes that we are aware of. Such a swing in origins is usual for December.
  • We are releasing a form_factors metric for the CrUX API and the CrUX History API. This metric is reported when formFactor is not specified as an input, that is, when requesting statistics for the overall origin or URL. In this case, we report the fraction of the page loads in the CrUX dataset that came from desktop, phone, and tablet—and in the CrUX History API we report a timeseries of these fractions. We think this data is often useful for further understanding the changes in overall performance of a site over time and for identifying opportunities. Note that in CrUX BigQuery, this information was already available (e.g. see [desktop|phone|tablet]Density in the metrics_summary table) and the CrUX Dashboard displays the desktop / phone / tablet fractions in its "Device Distribution" page.

Notable stats

202311

Announcement

Publication date

  • December 11, 2023

What's new

  • More improvements across most of the metrics this month, though this time we're not aware of any particular changes to attribute this to.
  • Our documentation has moved to new infrastructure, similar to the web.dev move in October. A huge benefit of this is that the CrUX documentation (and all other Chrome documentation) is available in 20 additional languages through machine translation, opening it up to many more people. Raise an issue if you spot any problems with this site move.

Notable stats

202310

Announcement

Publication date

  • November 14, 2023

What's new

Notable stats

202309

Announcement

Publication date

  • October 10, 2023

What's new

Notable stats

202308

Announcement

Publication date

  • September 12, 2023
What's new
We're seeing continual improvements across most of the metrics which seem to be driven by the following changes:
Note that all of these changes are still rolling out so the full effect of all these changes has not been seen yet and hopefully we will see further improvements next month too.
On a less positive note, INP pass rates remained static, showing it's a more tricky metric to optimize for.

Notable stats

202307

Announcement

Publication date

  • August 8, 2023

What's new

  • As noted last month, we made progress on origins with redirects on their root page not showing in CrUX and since then we resolved the last of the known issues, and this month, no origins should be excluded for this reason.
  • As warned in the last few releases, we have now removed the experimental prefixed Interaction to Next Paint metric from CrUX BigQuery, API, and History API. Users need to use the non-prefixed field to access this metric.

Notable stats

202306

Announcement

Publication date

  • July 11, 2023

What's new

  • We are pleased to announce that we have finally made progress on the long-standing root page redirect issue mentioned last month. Some origins which redirect their root page, and were therefore excluded from CrUX due to our lack of handling this properly, should now start to be included again from this month. Note that some origins are still being resolved, so not all origins are in this month's dataset. However, it is great that we have made some good progress on this now and we hope to have the issue fully resolved in the near future.
  • This is the last month that the INP metric will be available in the CrUX BigQuery, API, and History API both with and without the experimental prefix. We encourage users to move to the non-prefixed field, as the experimental prefix fields should now be considered deprecated and will be removed in 30 days.

Notable stats

202305

Announcement

Publication date

  • June 13, 2023

What's new

  • We're seeing a slight decrease again this month for LCP and FCP due to the final roll outs of the change in LCP to ignore low-entropy images and correction in paint timing discussed last month. However, despite this, the overall good LCP, FID, and CLS rates are up slightly from last month.
  • We are aware that some origins that redirect their root page (for example, https://www.example.com that automatically redirects to https://www.example.com/en/) are not showing origin-level data in CrUX, and so are also not appearing in the BigQuery dataset. We're also aware that this has been going on for some time. Unfortunately, this is proving a difficult issue to resolve, and so we still don't have an estimated time of when this will be fixed.

Notable stats

202304

Announcement

Publication date

  • May 10, 2023

What's new

Notable stats

202303

Announcement

Publication date

  • April 11, 2023

Notable stats

202302

Announcement

Publication date

  • March 14, 2023

What's new

  • We're seeing further improvements to the good INP rate which has increased by 2.3% to 77.3% of origins due to the improvement in the Chrome scheduler mentioned last month, which was still rolling out during February.

Notable stats

202301

Announcement

Publication date

  • February 14, 2023

What's new

  • Good INP increased by 4.3% to 75.6% of origins, primarily due to an improvement in the Chrome scheduler. The next frame is now scheduled with highest priority after discrete input events, which shortens the lag between input event and visual update.
  • This month we've launched a CrUX History API, which provides 25 weeks of historical data at both origin and URL level. We've written a post detailing how to use this new API, and a Colab showing you an example of how to plot this historical data in various graphs.

Notable stats

202212

Announcement

Publication date

  • January 10, 2023

What's new

  • No significant updates in this release.

Notable stats

202211

Announcement

Publication date

  • December 13, 2022

What's new

  • No significant updates in this release.

Notable stats

202210

Announcement

Publication date

  • November 8, 2022

What's new

  • In this release we have further broken down our "rank" popularity. Previously these were in groupings of log 10 (i.e. top 1,000 sites, top 10,000, top 100,000 sites, … up to top 100 million sites). From this release these are provided in half-rank steps (i.e. top 1,000 sites, top 5,000 sites, top 10,000, top 50,000 sites, top 100,000 sites, top 500,000 sites… up to top 100 million sites). The top rank (1,000) won't be altered with this change (i.e. we are not providing a top 500 rank).

Notable stats

202209

Announcement

Publication date

  • October 11, 2022

What's new

  • The CrUX API now includes collectionPeriod object showing what day's data is included in the response as shown. More details in the CrUX API docs.
    "collectionPeriod": {
      "firstDate": {
        "year": 2022,
        "month": 9,
        "day": 12
      },
      "lastDate": {
        "year": 2022,
        "month": 10,
        "day": 9
      }
    }

Notable stats

202208

Announcement

Publication date

  • September 13, 2022

Notable stats

202207

Announcement

Publication date

  • August 9, 2022

What's new

  • This release includes a change which records the CLS metric when the tab is backgrounded in addition to tab close. See the CLS changelog for details.
  • We have identified a bug in our origin filtering which unfortunately affects a small number of origins, including some popular ones, which are missing from this release. We hope to get this fixed for future releases.

Notable stats

202206

Announcement

Publication date

  • July 13, 2022

What's new

  • The number of origins covered by the CrUX dataset increases once again in this release due to the optional dimensions feature introduced in the previous release. We expect the size to be much more stable from here.

Notable stats

202205

Announcement

Publication date

  • June 14, 2022

What's new

  • The percentage of origins having good LCP experiences increased this month by 2.4%, thanks in large part to an improvement in the performance of Chrome on Android. This was a major contributing factor to the percentage of origins having good Core Web Vitals increasing by 2.1%.
  • With this release, CrUX records no longer require effective connection type and form factor, so we're including data for which these fields are NULL, indicating all effective connection types or form factors, respectively. This allows us to increase origin coverage by 28.2%.
  • An unrelated issue with our data pipeline prevents us from adding new origins in this release, so we're only including origins for which we published data in the previous 6 months. Due to this, experimental.popularity.rank has some gaps. For example, there are only 904 origins in the top 1,000. This is unique to the 202205 release and should be fixed next month.

Notable stats

  • 11,024,795 origins
  • 42% of origins have good Core Web Vitals

202204

Announcement

Publication date

  • May 3, 2022

What's new

  • This release includes experimental.interaction_to_next_paint, our updated responsiveness metric.
  • experimental_interaction_to_next_paint in the CrUX API, with thresholds 200ms and 500ms.
  • experimental_time_to_first_Byte in the CrUX API, with thresholds 800ms and 1800ms.

Notable stats

202203

Announcement

Publication date

  • April 12, 2022

What's new

  • No significant updates in this release.

Notable stats

202202

Announcement

Publication date

  • March 8, 2022

What's new

  • This release includes experimental.responsiveness, our candidate for the new responsiveness metric. See web.dev/blog/responsiveness for details.

Notable stats

202201

Announcement

Publication date

  • February 8, 2022

What's new

  • No significant updates in this release.

Notable stats

202112

Announcement

Publication date

  • January 11, 2022

What's new

  • No significant updates in this release.

Notable stats

202111

Announcement

Publication date

  • December 14, 2021

What's new

  • We've removed the old CLS metric: experimental.uncapped_cumulative_layout_shift is no longer in this BigQuery release, and CrUX API no longer serves experimental_uncapped_cumulative_layout_shift. Use the current CLS metric instead.
  • An LCP fix rolled out with M96.

Notable stats

202110

Announcement

Publication date

  • November 9, 2021

What's new

  • No significant updates in this release.

Notable stats

202109

Announcement

Publication date

  • October 12, 2021

What's new

  • There was a change to FCP in Chrome that ignores content with a style of opacity:0, increasing the user-perceived accuracy of the metric. Read more about it in the Chrome Speed change log.

Notable stats

202108

Announcement

Publication date

  • September 14, 2021

What's new

  • No significant updates in this release.

Notable stats

202107

Announcement

Publication date

  • August 10, 2021

What's new

  • No significant updates in this release.

Notable stats

202106

Announcement

Publication date

  • July 13, 2021

What's new

  • Changes to the FID metric took effect in Chrome 91, resulting in improved FID experiences involving double-tap-to-zoom on mobile.

Notable stats

202105

Announcement

Publication date

  • June 8, 2021

Notable stats

Notable Changes - The BigQuery update 202105 includes the same CLS and LCP updates that we released on June 1 for the CrUX API (see below). The previous CLS metric will be available for a limited time as experimental.uncapped_cumulative_layout_shift.

API Update
June 1, 2021

Notable Changes - The new Cumulative Layout Shift definition is now the default metric surfaced as cumulative_layout_shift, the previous Cumulative Layout Shift metric will be available for a limited time as it is phased out as experimental_uncapped_cumulative_layout_shift. - Largest Contentful Paint has undergone adjustments in recent Chrome versions and has been updated similarly in CrUX. - First Contentful Paint tri-binning thresholds have been updated to be: [0-1.8s], (1.8s-3s), [3s-∞].

202104

Announcement

Publication date

  • May 11, 2021

Notable stats

202103

Announcement

Publication date

  • April 13, 2021

Notable stats

202102

Announcement

Publication date

  • March 9, 2021

What's new

  • Coarse origin ranking data will be available in the CrUX dataset on BigQuery, starting in the February 2021 (202102) release. With this feature, we'll be able to unlock new insights into the aggregate user experiences of the head, torso, and tail of the web. All origins in the global and per-country tables will be assigned rank magnitude values under the experimental namespace, experimental.popularity.rank. Values will be powers of 10, starting with the top 1,000, then 10,000, 100,000... For example, a rank magnitude of 10,000 means that an origin is among the top 10,000 most visited origins in the dataset (learn more).
  • Tell us what you think of this new feature. We welcome your feedback on the CrUX discussion forum.

Notable stats

202101

Announcement

Publication date

  • February 10, 2021

Notable stats

202012

Announcement

Publication date

  • January 12, 2021

Notable stats

202011

Announcement

Publication date

  • December 8, 2020

What's new

Notable stats

202010

Announcement

Publication date

  • November 10, 2020

What's new

Notable stats

202009

Announcement

Publication date

  • October 13, 2020

What's new

  • No major changes in this release!

Notable stats

202008

Announcement

Publication date

  • September 9, 2020

What's new

  • There was a bump in origin coverage this month (+3.3%) due in part to a data pipeline change. Previously, for a user experience record to be included in the dataset, it must have included FP, DCL, and OL measurements. Now, the only required metric is FCP. This change helps to ensure that we're capturing more meaningful user experiences.

Notable stats

202007

Announcement

Publication date

  • August 11, 2020

What's new

  • No major changes in this release!

Notable stats

202006

Announcement

Publication date

  • July 14, 2020

What's new

  • User experiences served from Google's AMP cache are now attributed to the publisher's origin in CrUX. Effects will vary depending on the amount of AMP cache traffic relative to the rest of the website, but most origins see less than a 5 percentage point change in Core Web Vitals performance, with most effects being positive.

Notable stats

202005

Announcement

Publication date

  • June 9, 2020

What's new

  • There were two LCP updates, which should have a minimal effect on most origins
  • There was one FID update, which may result in some sites observing faster FID experiences

Notable stats

202004

Announcement

Publication date

  • May 12, 2020

What's new

  • There was a small improvement to the way LCP is measured in M81. This may manifest itself in your CrUX data as a slight shift in slower LCP or in rare cases too few LCP samples to include for your origin.
  • Introducing Web Vitals! CrUX is committed to providing you with all of your Core Web Vital metrics to give you an accurate picture of how real users experience the web.
  • Do you query data on a per-country basis? Give the new experimental.country table a try! It's optimized to make querying across country and monthly datasets easier.

Notable stats

  • 6,389,861 origins

202003

Announcement

Publication date

  • April 14, 2020

What's new

  • CLS histogram start and end values changed from INT64 to NUMERIC type and divided by 100. This change aligns CLS with the fractional layout shift values provided by the Layout Instability API.

Notable stats

  • 5,937,298 origins

202002

Publication date

  • March 10, 2020

Notable stats

  • 6,366,736 origins

202001

Publication date

  • February 11, 2020

What's new

Notable stats

  • 5,976,293 origins

201912

Notable stats

  • 5,532,155 origins

201911

Notable stats

  • 5,821,306 origins

What's new

  • The FID metric was moved from experimental.first_input_delay to first_input.delay
  • The CLS metric was moved from experimental.cumulative_layout_shift to layout_instability.cumulative_layout_shift

201910

Notable stats

  • 5,752,729 origins

201909

Notable stats

  • 6,008,004 origins

What's new

  • The LCP metric was launched as largest_contentful_paint
  • CLS was updated to take move distance into account. Coverage may be lower while Chrome users upgrade to the latest version of the Layout Instability API

201908

Notable stats

  • 6,011,463 origins

What's new

  • FID coverage has returned to normal
  • The average percent of fast experiences for most metrics dropped by about 2%, this appears to be due to a bug in Chrome

201907

Notable stats

  • 5,612,504 origins

What's new

  • There was an incremental update to Chrome's FID implementation, which included pointer events on mobile. Coverage will be lower while Chrome users update to the latest version

201906

Notable stats

  • 5,624,797 origins

What's new

  • The TTFB metric was added to the list of experimental metrics as experimental.time_to_first_byte

201905

Notable stats

  • 5,884,155 origins

What's new

  • The CLS metric was added to the list of experimental metrics as experimental.cumulative_layout_shift

201904

Notable stats

  • 5,744,982 origins

What's new

  • There was an incremental update to Chrome's FID implementation. Coverage will be lower while Chrome users update to the latest version

201903

Notable stats

  • 5,703,255 origins

What's new

  • There was an incremental update to Chrome's FID implementation. Coverage will be lower while Chrome users update to the latest version

201902

Notable stats

  • 5,464,560 origins

201901

Notable stats

  • 5,351,287 origins

201812

Notable stats

  • 4,654,112 origins

201811

Notable stats

  • 4,697,003 origins

201810

Notable stats

  • 4,374,729 origins

201809

Notable stats

  • 4,375,805 origins

201808

Notable stats

  • 4,386,422 origins

What's new

  • Histogram bins have been normalized to consistent widths (more info)

201807

Notable stats

  • 4,202,945 origins

201806

Notable stats

  • 4,134,123 origins

What's new

  • The FID metric was added to the list of experimental metrics as experimental.first_input_delay (learn more)

201805

Notable stats

  • 4,162,633 origins

201804

Notable stats

  • 3,970,181 origins

201803

Notable stats

  • 3,589,954 origins

201802

Notable stats

  • 3,237,524 origins

201801

Notable stats

  • 3,086,603 origins

What's new

201712

Notable stats

  • 1,939,945 origins

201711

Notable stats

  • 1,237,407 origins

What's new

201710

Notable stats

  • 10,000 origins

What's new