-
You’re right! It does seem a bit odd. But actually, there might be some reasons why there’s a slight discrepancy. For one, GA4’s pre-made reports aren’t always super reliable. So we often suggest using the Explorer or exporting the data for analysis somewhere else if that’s a possibility for you.
Secondly, you’ll want to double-check you don’t have Google Signals enabled because this alters the thresholding and sampling logic. Also, switching to device-only reporting can help with thresholding issues. Just a side note, once Google Signals is enabled, it seems the thresholding logic can’t be fixed even if you disable them afterwards.
Another thing to consider is Event Properties Cardinality. Despite Google’s claims, GA4 still has a fair share of bugs and less-than-ideal features. The cardinality of your event property values is one example. If it’s too high, sampling starts becoming a bigger issue and a larger percentage of your events comes up as ‘Other(Other)’, and this happens even if you’re not using the high cardinality dimension in your report.
There are also data retention limitations. Unfortunately, GA4 doesn’t provide free access to old data anymore, so if you’re trying to count old events, that may be a problem. UA does let you see old events, but GA4 does not.
These suggestions are quite broad. For a better understanding, you might need to do further problem-solving on your end to find out exactly which datapoints and at what times aren’t getting counted. Exporting data to BigQuery might aid in narrowing things down. Right now, many analysts tend to avoid directly comparing GA4’s data to that of UA – even though knowing the difference can still be useful. But it’s kind of become the norm in the industry today.
