Hi all,
We’re in the process of migrating from GA4 to Piwik PRO Analytics (PP), but still running GA4 and PP side by side to validate the data. We’re obviously not expecting a 100 % match in the data between the platforms, but so far we’ve only gotten 40-50 % of the session volume in PP that we get in GA4, and we’re starting to believe the issue is in PP’s server endpoint(s). Page view hits have a smaller discrepancy, but even those are around 30-40 % lower than what’s reported by GA4.
We have a setup with client-side GTM running PP tags, which send hits to server-side GTM to a PP Client and a PP tag to process them. I.e. PP Tag Manager is not in use at all.
The client-side PP Basic Tracking Code (BTC) disables injection of the tracking client and loads it from our server (we modified the tag template to do this, instead of a custom HTML tag like in the support page here). This works fine, and we even tested the setup by sending the hits straight to PP (circumventing our sGTM), and the match rate with GA4 was much better, like 90-100 %.
There are no augmentations on our sGTM that would treat GA4 differently from PP. From our GCP server the requests seem to pass on to Piwik fine, and the Piwik endpoint seems to respond fine to normal debugging. We’ve also looked at our server logs and there are very few errors, not nearly enough to explain the difference, so Cloud Run is not the bottleneck.
But when we look at sessions in time, lost session rate drops (i.e. is significantly lower) when session counts are higher. This might indicate that there’s an issue in Piwik’s endpoint processing.
We’re in the process of discussing this with Piwik support, but since we’ve been at it for months already, I wanted to check if anybody else would have a similar setup and might have encountered something similar, or maybe gotten a setup like this working.
