The 5 most common problems we’ve seen with GA4 migrations

Eugh final panic. Barely two weeks left and your GA4 numbers still aren’t lining up with your UA numbers.

Here are the most common things we've seen fail.

We've tried to provide screenshots/data examples where possible, but it hasn't been possible in all the cases.

Does your cookie banner respect analytics?

What’s the issue:

The cookie banner was configured one way for UA and another way for GA4. i.e. perhaps GA4 was ignoring the cookie banner and firing anyway, or vice versa.

This is the most common issue we’ve seen.

What does it look like in the data:

• You've got more/less sessions. You can see it to a lesser extent in pageviews.
• The easiest place to see it is in a country by country breakdown of sessions, you can see it in the absolute numbers.
  • If you breakdown traffic by country split across country you’ll often see the difference in EU countries vs non EU countries. (see below).

What is the absolute vs relative difference here?

Brief sidebar as we're about to mention this a bit.

• Absolute: is the % difference in traffic between a segment of UA and GA4.
• Relative: is the % difference of total traffic of a segment in GA4 to the % of total traffic of that segment in UA.
  • I.e. USA traffic is 40% of total GA4 and 30% of total UA traffic, so relative is 10%.

Relative is particularly helpful for sessions and users as it helps accounts for things like the session model changed from UA to GA4 so all the absolute numbers will be different a certain degree.

What to do?

Test with dev tools or the GA debugger, confirm and fix.

(Sorry we haven't got space in this post to go through the details of how to fully debug GA4, that's deserving of it's own post.)

Are you counting client side page views twice?

What’s the issue:

Client side page views aren't being tracking correctly and now they are. (Or very occasionally vice versa.)

i.e. Suppose you have a shop with a faceted navigation where each filter, sends a client side call to change the data but doesn’t reload the page. We’ve seen quite a few cases where this didn’t track in UA but is tracking in GA4.

This is the second most common issue we’ve seen.

What does it look like in the data?

• You’ve got more relatively more pageviews, wherever the client side interactions are (typically category pages, search pages etc.)

What to do?

Test with dev tools or the GA debugger, confirm and fix.

Are you seeing a difference between LC session & LNDC session attribution

What’s the issue:

We had a mistake here that made it through edit which was spotted by a sharp eyed Martin Winberg. We didn't mean the interface we meant the GA4 BQ export.

The GA4 BigQuery sources you see at event level are session level by default and don't use the same attribution as UA.

It uses the last click attribution rather than last non direct click attribution.

(This is separate from other attribution change you can't do anything about with: cif there are multiple sources in a session window, rather than splitting now it will use the first.)

What does it look like in the data?

• A relative increase in (none) for your medium/channel grouping attribution at the session level.
• We've usually seen this be somewhere in the range from a 5-14% change where (none) has increased relatively by that much.

What to do?

You can configure a LNDC model with the GA4 BigQuery export, but it's relatively complicated if you're not comfortable with data modelling (which is what you’re seeing in the comparison screenshots).

At the very least set-up your GA4 BigQuery export as even if you can’t do it now, someone could do it for your in the future, it costs very little and it’s not retroactive.

Are you firing multiple GA4 tags?

What’s the issue:

Multiple versions of GA4 added and duplicating hits.

What does it look like in the data?

• Notably higher pageviews (usually 30% more than UA) often across the board. (Although sometimes not).

What to do?

Check your tag manager isn’t firing multiple hits. Otherwise use dev tools and look at the request initiator to try and understand where they’re coming from.

Are you accidentally session splitting in UA?

What’s the issue: UTM parameters on internal links caused session splitting and moving to GA4 has caused the splitting to stop.

What does it look like in the data?

A decrease in sessions which are happening at all times of the day (as opposed to just over midnight).

What to do?

Nothing unfortunately. The UA was set-up badly and now it’s not. Not much you can do with that :(

A little less likely

There’s also one relatively un-common issue, but we’ve seen it sometimes with more enterprise tool sets:

• Setting a non-unique custom session ID in GA4
  • This usually only happens if you’re doing some relatively heavy customisation, but its worth checking to make sure that value is a unique value as opposed to something like a hash of browser + device (which we've seen.)

If you have exported your data to BigQuery then the following query will let you check:

Two weeks left

Good luck all. Of we all go into the new world.

If you’re still running at this point, hopefully it goes well and if you do want any help with getting better reporting from GA4 please get in touch!