Introducing historical web performance data via the CrUX History API

37 views
Skip to first unread message

Barry Pollard

unread,
Feb 7, 2023, 4:25:45 PM2/7/23
to chrome-ux-re...@chromium.org
Hi CrUX users,

We are pleased to announce a new feature to the Chrome User Experience Report: the CrUX History API.
https://developer.chrome.com/blog/chrome-ux-report-history-api/

This API gives 6 months of historical CrUX API data to help you understand the web performance trends of your website. Like the daily API, the History API is available to query for both origins and individual URLs that meet the eligibility criteria.

The above blog post gives examples on how to query the data, and also includes a demo Colab which uses the API to chart the history of Core Web Vitals (CLS, FID, LCP), and additional metrics (TTFB, FCP, INP) for a site or URL.
https://colab.sandbox.google.com/github/GoogleChrome/CrUX/blob/main/colab/crux-history-api.ipynb

For even more details, the technical specifications have been added to our documentation: https://developer.chrome.com/docs/crux/history-api/

We're eager to find out how useful this API will be to you, and what insights you end up obtaining from it. So do let us know!

Barry

Barry Pollard

unread,
Feb 8, 2023, 11:06:54 AM2/8/23
to Walter Lee, Chrome UX Report (Discussions), 'Barry Pollard' via Chrome UX Report (Announcements)
Hi Walter,

This was discussed a lot during the design phase and we elected explicitly NOT to do this, for a number of reasons:
- The 28-day period was chosen as being a better representative of the site's actual performance - less subject to daily variances, or short spikes in traffic.
- It's confusing to have two different time periods.
- This would likely lead to a lot of questions "Why do I get X when I look at the weekly numbers, but Y when looking at the monthly ones".
- It would likely result in a lot more entries not being able to be returned as they may not meet the eligibility criteria.

We recommend using CrUX as good (stable) indicators of a website's general performance, but we do think it needs to be supplemented with RUM measurements as well to allow more deeper investigations that can be broken down in various ways (different times, per page...etc.) as discussed here: https://web.dev/crux-and-rum-differences/#benefits-of-supplementing-crux-with-a-rum-solution

Thanks,
Barry


On Wed, 8 Feb 2023 at 15:58, Walter Lee <happy...@gmail.com> wrote:
Hi, Barry,

Just tried it and all great work ! Thank you! One suggestion, can we have a NEW option just to do collection period = 1 week instead of current  previous 28-days aggregated data please , reason : we can easily see if any significant drop in web perf if we can just do 7 days for last week, e.g. a new release added blocking js, then it will drop 10% in LCP if we just look at 1 week data instead of 4 week rollup now (only ~2.5% drop shown ? ) 

Walter
Reply all
Reply to author
Forward
0 new messages