Cookie Consent by Free Privacy Policy Generator

The Best Of

Go to the Best Of the SEO Community.

Noah
Noah
Jan 19, 2025, 10:12 AM
Forwarded from another channel:
Forwarded thread from another channel:
David Gutierrez
David Gutierrez
Dec 11, 2024, 7:49 AM
Can you help me wrap my head around something? If I look at data in smaller chunks, my total users is greater than new users, which is correct. But if I stretch that to the last 12 months, new users exceeds total users, yet there is no instance on a particular date that new is actually exceeding total.
Is this an attribution issue, where the new user is getting attributed to multiple channels or something, thus duplicating their entry? Thanks for any help!
Nico Brooks
Nico Brooks
Dec 11, 2024, 10:00 AM
That's certainly weird. It makes sense to me that the New users count would approach Total users the further back you go, given any tracking platform's inability to accurately identify a human being over time. So, let's say that a 12-month view of New users equalling Total users is expected. Then perhaps the fact that you are seeing _more_ New users is rearing it's ugly head? The difference you are seeing is well within the HLL++ margin of error.
David Gutierrez
David Gutierrez
Dec 11, 2024, 10:24 AM
Was hoping you'd reply :)
yeah good point - the larger the date range, the high proportion we'd likely see of total users being new users as well.
Dana DiTomaso
Dana DiTomaso
Dec 12, 2024, 11:19 AM
How many first_visit events do you have in this timespan?
David Gutierrez
David Gutierrez
Dec 12, 2024, 11:21 AM
the exact same amount of new users. Just checked.
Dana DiTomaso
Dana DiTomaso
Dec 12, 2024, 11:25 AM
Yeah so that is GA4's real interpretation of how many new visitors you had, since first_visit only records if the cookie isn't found. HLL++ only applies to session counts, not users (as far as I know?), so I don't think that is the issue. I'm just trying to wrap my mind around the total vs new in the organic search row.
David Gutierrez
David Gutierrez
Dec 12, 2024, 11:26 AM
referral has more new too.
Yeah I can't wrap my head around it.
Dana DiTomaso
Dana DiTomaso
Dec 12, 2024, 11:27 AM
Is this for a worldwide business? Sometimes sessions crossing midnight can cause extra first_visits
Dana DiTomaso
Dana DiTomaso
Dec 12, 2024, 11:30 AM
Like if someone is a first_visit at 11:59pm and then when it crosses midnight, same client ID so still one user, but a new first_visit would be recorded (to my understanding).
David Gutierrez
David Gutierrez
Dec 12, 2024, 11:38 AM
why would it fire twice if the event has already been recorded once?
Dana DiTomaso
Dana DiTomaso
Dec 12, 2024, 11:50 AM
I bet that's it then. Google doesn't have a lot of documentation on the whole why around it. In UA, the session would totally restart at midnight, that doesn't happen in GA4. But to my understanding, you'll end up with two first_visit events. You might be able to dig into the data in BigQuery and find it happening if you have it there. But I'd say with that kind of global audience and the volume of first vs total, that absolutely looks like the culprit.
Dana DiTomaso
Dana DiTomaso
Dec 12, 2024, 11:50 AM
It's something that I have a note to myself to do more research on, I also see it on global sites.
Nico Brooks
Nico Brooks
Dec 12, 2024, 12:12 PM
The fact that first_visit event count and New users is exactly the same is really interesting. I assumed that New users was a unique count of pseudo_user_id or user_id with first_visit events, but it does seem like it's just a count of first_visits. Google _does_ , and based on a precision of 14, there's a at a confidence level of 95%. This jibes with what I see when I compare Total and Active user metrics I generate from BigQuery without HLL++ to metrics in GA4.
So, maybe New users is squirrely because of the midnight issue, but Total users is _definitely_ squirrely because of HLL++.
Nico Brooks
Nico Brooks
Dec 12, 2024, 12:14 PM
The other thing that occurred to me - if Google Consent Mode is enabled and modeling is working, that introduces yet another layer of estimation.
David Gutierrez
David Gutierrez
Dec 12, 2024, 12:27 PM
I'm pretty sure they are not setup for appropriate consent / GDPR / cookies, etc.
Nico Brooks
Nico Brooks
Dec 12, 2024, 1:08 PM
Ok, so you have that to look forward to ????
Dana DiTomaso
Dana DiTomaso
Dec 12, 2024, 1:36 PM
What I was told by the Googs was that HLL was really mostly used for sessions, not users because they use event counts for users
Dana DiTomaso
Dana DiTomaso
Dec 12, 2024, 1:36 PM
Even though that disagrees with their documentation!
Dana DiTomaso
Dana DiTomaso
Dec 12, 2024, 1:36 PM
But who knows, classic Google
Dana DiTomaso
Dana DiTomaso
Dec 12, 2024, 1:37 PM
Either way for sure, in every site I've reviewed, new users matches first_visit
Nico Brooks
Nico Brooks
Dec 12, 2024, 1:39 PM
I can see how that would work for New users because of the first_visit event, but what events would be counted for Total users and Active users?
Dana DiTomaso
Dana DiTomaso
Dec 12, 2024, 2:56 PM
I'm sure there is some sort of HLLing going on, despite what that Googler said (I wish I could recall who it was, it was on a call).
Nico Brooks
Nico Brooks
Dec 12, 2024, 2:59 PM
Well, once told me she focused on trends versus absolute numbers, because Analytics can't really identify people anyway. I have a tough time accepting it, but it was good advice.
Dana DiTomaso
Dana DiTomaso
Dec 12, 2024, 3:14 PM
Ha! Who would that have been?

Our Values

What we believe in

Building friendships

Kindness

Giving

Elevating others

Creating Signal

Discussing ideas respectfully

What has no home here

Diminishing others

Gatekeeping

Taking without giving back

Spamming others

Arguing

Selling links and guest posts


Sign up for our Newsletter

Join our mailing list for updates

By signing up, you agree to our Privacy Policy and Terms of Service. We may send you occasional newsletters and promotional emails about our products and services. You can opt-out at any time.

Apply now to join our amazing community.

Powered by MODXModx Logo
the blazing fast + secure open source CMS.