Details on which page views are considered in CrUX data

[Kristian Sköld]

Hi team, 

we are accelerating a number of customer sites and during the discussion with one of our customers a question came up, which I couldn't answer by looking into the various documentation available.

Does Chrome send the data of all page views to the CrUX data set or only the first page view in a session?

I'm certain that all the page views are included in the data, but our customer is convinced it is only the first.

In https://groups.google.com/a/chromium.org/g/chrome-ux-report/c/eMBEeg-qdsk/m/38gx1BTzAAAJ Boris states, that the whole User Experience is collected - but this is a discussion regarding SPAs and could still be interpreted as SPA having only 1 full page view anyway.

In https://groups.google.com/a/chromium.org/g/chrome-ux-report/c/SYWzGqCsNpw/m/oYrUPgfVDAAJ Rick explains:

1. Do Chrome UX reports show 100% of the users interactions with sites?

Per the CrUX docs, only the experiences "from users who have opted-in to syncing their browsing history, have not set up a Sync passphrase, and have usage statistic reporting enabled" are included in the report. Of those experiences, there isn't any other downsampling.

The quote from the docs explains, which users are included in the data set, but it is not clear whether "Of those experiences, there isn't any other downsampling."  is just referring to no downsampling in regards of users or also in regards to page views/navigations. What exactly does "experiences" include - technically.

A more detailed definition would be really helpful.

Is there maybe also a way to see exactly which data is being sent by the Chrome browser while navigating a site? These data requests are of course not included in the dev tools network tab, but maybe there is another way to access them locally.

Thanks a lot in advance for your help here.

The CrUX data and Web Vitals are such powerful tools and extremely helpful in raising awareness and understanding regarding the performance topic - but there are still also a number of confusing topics, which are hard to explain without having some official statement from Google :)

[Rick Viscomi]

Hi Kristian,

> Does Chrome send the data of all page views to the CrUX data set or only the first page view in a session?

At the origin level, CrUX includes all data from all page views from eligible users (opt-in, etc). As you alluded, things get messy with SPAs particularly, because these websites are navigating pages in ways that break the standard web platform way of measuring user experiences. So in effect CrUX is blind to these "soft navigations" from SPAs and the entire multi-page experience ends up getting attributed to the initial page view. This is not ideal and multiple groups are tackling this problem from both the metrics and platform API standards perspectives. For example, Evolving the CLS metric is our response to the challenges of measuring CLS in these long-lived apps.

> The quote from the docs explains, which users are included in the data set, but it is not clear whether "Of those experiences, there isn't any other downsampling."  is just referring to no downsampling in regards of users or also in regards to page views/navigations. What exactly does "experiences" include - technically.

There's no downsampling of any kind. By that I mean we use *all* of the data points from eligible users that come in when calculating the aggregate metrics.

An "experience" in this context refers to the metrics collected during a user's page view, with the caveat above of some page views bleeding into each other in SPAs.

> Is there maybe also a way to see exactly which data is being sent by the Chrome browser while navigating a site? These data requests are of course not included in the dev tools network tab, but maybe there is another way to access them locally.

CrUX data is based on the open source URL-Keyed Metrics (UKM) API. I've never tried it but according to these docs it should be possible to monitor UKM data on the chrome://ukm page. It may involve building Chrome from source though :-/ (Has anyone monitored their UKM data before?)

> The CrUX data and Web Vitals are such powerful tools and extremely helpful in raising awareness and understanding regarding the performance topic - but there are still also a number of confusing topics, which are hard to explain without having some official statement from Google :)

Glad to hear you've found these tools useful and I hope this helped.

Rick

[Ajendra Singh]

Hi,

I am trying to figure out what is the exact way of calculating origin-level CrUX report ?
I see that 80% of the pages/URLs are flagged as passed individually on the origin, but still my origin-level CrUX data is flagged as failed. What could be the reason for this and am I missing something here ?

Thanks in advance, hope to hear from you soon

[Barry Pollard]

Hi Ajendra,

It is best to open a new discussion rather than reply to an older, seemingly unrelated one.

In that you should give details of where you are getting your information (possibly with screenshots). How do you know that 80%of the pages are passing?

In addition the origin-level score is based on all page views. So it’s entirely possible you have 80% of your pages all passing, but the 20% that are failing are visited much, much more often by your site’s visitors so they contribute more to your origin score.

Thanks,

Barry

'Disclaimer- Any content of this e-mail and files transmitted along with it, may contain confidential and privileged information of upGrad Education Pvt. Ltd. ("Upgrad") and are for the sole use of the addressee/s indicated in this email. All intellectual property rights in the information shall belong to Upgrad or its respective owners in perpetuity. If you are not the indicated addressee/s and or have received this e-mail in an error or through other phishing or unauthorized sources, kindly notify the sender by replying to this e-mail immediately. Please destroy this e-mail and any attachments permanently in that case. You are hereby notified that any dissemination, distribution, copying or unauthorized use of this message and any attachment herewith, is strictly prohibited. Opinions, conclusions, and other information in this message that do not relate to the official business of Upgrad shall be understood as neither given nor endorsed by it. Unless asked specifically in writing from Upgrad, you shall not share any personally identifiable information/ documents over the email and Upgrad takes no responsibility of such personally identifiable information shared by you. In the event such information is sought, you shall ensure that the same is shared in an encrypted form and not otherwise. If you are in receipt of any personally identifiable information from Upgrad, you shall comply with the relevant applicable data protection laws and regulations including but not limited to General Data Protection Regulations and shall process such information only as per the relevant applicable data protection laws.'

--
You received this message because you are subscribed to the Google Groups "Chrome UX Report (Discussions)" group.
To unsubscribe from this group and stop receiving emails from it, send an email to chrome-ux-repo...@chromium.org.
To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/chrome-ux-report/efaae4ef-4853-45d1-b01c-0418c2d1183cn%40chromium.org.

[Ajendra Singh]

Thanks for the quick response Barry and apologies for replying to this discussion, will keep this in mind going forward.

I believe those 20% pages are the least visited ones and that is why they don't even show the CrUX dataset for that particular URL, and instead show the origin-level CrUX data on PSI. Let me analyse this using APIs exposed and the page-views on each URL and come back on this by opening a new discussion.

Thanks again !