URL level, but no origin level

101 views
Skip to first unread message

Dave Smart

unread,
Feb 15, 2023, 12:36:58 PM2/15/23
to Chrome UX Report (Discussions)
Would it be normal for there to be URL level data available, but nothing for the origin?


I'm not totally surprised there's no origin data, the site has traditionally skirted in and out of having enough data, but it feels weird to have URL level for one URL, but no origin.

Dave Smart

unread,
Feb 15, 2023, 12:37:41 PM2/15/23
to Chrome UX Report (Discussions), Dave Smart
Should add, this only has url level for Desktop.

❄ Johannes Henkel

unread,
Feb 15, 2023, 4:36:55 PM2/15/23
to Dave Smart, Chrome UX Report (Discussions)
Yes. This is an edge case but expected to happen sometimes.

Currently we have origin level data for https://tamethebots.com, but it's OVERALL, that is, DESKTOP and PHONE and TABLET combined (any form factor). For simplicity's sake, below I'll assume there's no TABLET.
So, in BigQuery you'll see a row with a NULL value for form_factor.name - that's OVERALL. I tried this:
select * from chrome-ux-report.all.202301 where origin = 'https://tamethebots.com'
You can try a query against our APIs (the daily and the history), if you leave out the form factor parameter, you'll get origin level data for this.

But, if you're looking at this in PageSpeed Insights (PSI), you won't see origin data because we don't have data that distinguishes DESKTOP and/or PHONE, and PSI thus far doesn't display OVERALL (any form factor) data.

Now like you're saying, for https://tamethebots.com/tools/robotstxt-checker, we have DESKTOP data, but not PHONE. Reason being, for PHONE by itself, it's not sufficiently popular. You may notice also that for DESKTOP and OVERALL, we report the same histogram tri bins. This is intentional, otherwise you could get PHONE by subtraction, but the difference with the extra PHONE traffic is small enough that we're cool with equating DESKTOP and OVERALL, usually it won't affect your analyses.

Now, for the origin https://tamethebots.com, there's a bit more PHONE traffic, but still not enough to make it so that we can report the data - it's still below our "sufficiently popular" threshold. But now we run into another problem: If we again report DESKTOP to be the same as OVERALL for the origin, it'd be too distorted / misleading due to the extra PHONE data in OVERALL. So, we only report OVERALL for the origin.

--
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/9bccf7ec-59eb-40bc-ad58-1909b0b588fdn%40chromium.org.


--
johannes :-)

Dave Smart

unread,
Feb 15, 2023, 4:43:20 PM2/15/23
to Chrome UX Report (Discussions), joha...@google.com, Chrome UX Report (Discussions), Dave Smart
Hey Johannes

Thank you for that great explanation! Makes sense

Thanks
Dave 

Sofien Ben Ayed

unread,
Mar 23, 2023, 5:59:03 AM3/23/23
to Chrome UX Report (Discussions), Dave Smart, joha...@google.com, Chrome UX Report (Discussions)
Hey there!

I recently came across the same behavior while browsing through several websites. I noticed that even though I was targeting the same URL, I could only access URL-level data, not origin-level data.
I read Johannes' answer to a similar problem (huge shoutout for the detailed explanation!) and tried to see if I was facing the same edge case.
So, I targeted the URL https://www.luisaviaroma.com/it-it/:
  • URL-level data on DESKTOP/PHONE/OVERALL: YES
  • URL-level data on TABLET: NO
  • Origin-level data on DESKTOP/PHONE/TABLET/OVERALL: NO
According to what I understood, if this was the same edge case as https://tamethebots.com, origin-level data should be available when targeting overall data (<=> without specifying a formFactor).
But in this case, no origin-level data of any kind can be retrieved.

Would you happen to have any idea how this can happen? Let me know your thoughts!

Thank you,
Sofien
Reply all
Reply to author
Forward
0 new messages