We just rolled out Matomo (which is awesome). We noticed the same thing as this user:
The problem: In the “pages” report (and also the “entry pages” and “exit pages” reports) all URLs are mixed up independent on which sub domain they are running. First this is confusing, and second it seems to create wrong data! All index pages of the root and the sub domains are listed together as page URL “/index”. When I open the page overlay, the index page of the sub domain with the highest number of pageviews is displayed. But the data shown left hand in the “main metrics” section is the sum of all sub domains (pageviews, visits etc.). When I change the page overlay’s URL manually to another sub domain’s index page, the “main metrics” data stays the same.
So for example, we have URLs like this:
In Matomo > Behavior > Pages it shows art > /index which is actually this URL:
Which is incorrect and unexpected.
We're already running this in our tracking code:
_paq.push(["setDocumentTitle", document.domain + "/" + document.title]);
But that doesn't seem to apply to Pages and other reports.
I searched around and I don't think this user put in a ticket so I thought I'd create one.
Thanks for the report @brenc It is indeed problematic and not good enough as it is now, as it's confusing and not clear, and many users like you don't expect this behavior. The issue is that some users want the behavior of considering all alias URLs as pointing to the same site (and merging them together), while others (like you) prefer to keep them separate.
Maybe this is something we could address in Matomo 4.x?
Not needed for Matomo 4. This can be done any time later.
Have a same issue with mixing of page urls which become from all subdomains, but they are logically separated from "main" and other domains, and may have different pages. What I can do at this moment to separate page urls? Maybe just add a "subdomain" to page url variable (like in Google Analytics) or something like this? I can write own code if it needed, but want to get some info about where to start with that problem.
@diosmosis not sure I fully understand the PR correct but thinking as it's done in archiving, might need to add a comment that this will cause maybe issues when looking at week, month, year, ... that some will have the host while some rows won't and it'll be all mixed maybe?
not sure I fully understand the PR correct but thinking as it's done in archiving, might need to add a comment that this will cause maybe issues when looking at week, month, year, ... that some will have the host while some rows won't and it'll be all mixed maybe?
yeah, that would be an issue. the easiest way around this might be to make the setting unchangeable. otherwise, if we see both types when aggregating we can group by host in the one that includes host, then aggregate w/ the one w/ no host. there's no way to go from no-host to host though so that's the best we can do I think.