Search code examples
pagespeed-insightschrome-ux-report

CrUX dashboard looses all data - why?


Two domains are affected by disappearing of data from CrUX dashboard:

These domains (origins) are affected: this and that. Both domains are available for bot crawling, earn much traffic and in the last year didn't experienced any changes. Both domains have empty CrUX dashboards since Juni/July 2023.

I tried to build CrUX dashboards from scratch - and got the same situation: like would the data since Juni/Juli 2023 no longer exist.

I found some descriptions of the same issue, but without any cause description.

As usual cause of no data are mostly two things mentioned: unsufficient traffic and prevented visiting by special bots. Both causes I can't count as applicable.

Does somebody know something about causes of the CrUX data lost?


Solution

  • I used the following query on CrUX's BigQuery dataset

    SELECT date, rank, device
    FROM `chrome-ux-report.materialized.device_summary`
    WHERE date >= "2023-01-01" AND origin = 'https://mediaworx.com'
    ORDER BY date DESC
    

    Which shows the following:

    date rank device
    2023-11-01 50,000,000 null
    2023-10-01 50,000,000 null 
    2023-09-01 50,000,000 null
    2023-08-01 50,000,000 null
    2023-07-01 50,000,000 null
    2023-06-01 50,000,000 desktop
    2023-05-01 50,000,000 null
    2023-03-01 10,000,000 null
    2023-02-01 50,000,000 null
    2023-01-01 50,000,000 null

    This shows me a few things:

    • This origin is in CrUX
    • It is not split by mobile and desktop, as is required for the CrUX Dashboard, which is due to insufficient traffic in each.
    • The rank shows this is in the top 50 million most popular websites in the world as measured by Chrome browsers. This is the final rank bucket (CrUX actually only has 17,323,447 origins as of December 2023) so it's not unsurprising that there is insufficient traffic for mobile/desktop split and and this is commonly the case for sites in the 10 million, and 50 million rank.
    • There was not sufficient traffic to include the december 2023 data, even when not segmented by device.

    You say both domains have sufficient traffic but that does not appear to be the case for this one. Note that it requires Chrome users who are logged in and consented to sharing anonymous data with Google as detailed in the methodology, so if getting a lot of users from other browsers (or iOS) then these would not be included.

    The other origin (http://www.schoeffel.com) may not be the same issue. It was in the top 500,000 websites until last August and then just dropped off. This suggests that CrUX thinks it shouldn't be listed. I note the base origin (http://www.schoeffel.com) redirects with a 302 to the language specific one (http://www.schoeffel.com/de/de) so I wonder if that is causing problems somehow? CrUX used to have issues with handling such redirects of the base page. Was anything changed in here in August/September 2023?