Should GA4 be your main or only analytics choice? Google says yes; I don't buy it.
Let’s explore why.
When GA 4 was officially launched on October 14, 2020, Google said it would become the default choice when a new GA property was set up.
As of October 22, 2020, that has now become true. Here’s what you see when you set up a new property:
You can still choose to set up the older “Universal Analytics” (UA) version. Click on “advanced options” to toggle on the UA option, and get Google’s warning that this is not recommended for new Analytics users.
Seriously? That’s a strong warning. Let’s break that down to see why I totally disagree with it.
If you have an iOS or Android app, you won’t be able to track it without installing GA 4. I totally agree with that.
But if your company doesn’t have an app – which is a huge number of companies – giving up the app measurement is no big deal.
If you do have an app, then you need to at least install GA 4 as supplementary analytics.
Less is More
Google says the old version of GA had too many reports and was too overwhelming for the casual user. I can’t argue with that.
But can you get enough insight into user behavior just from the GA 4 user interface (UI)? I don’t think so. If you need to dive down into more details than what the UI offers, you have four choices:
- Choose one of the templates in the Analysis section
- Build your own template – the tools are relatively rich and if you’ve worked in Google Data Studio (GDS), you’ll get it fairly quickly
- Export the data to BigQuery and do your analysis there
- Build your reports in GDS
Basically, the complexity that leads to insights is outside of the UI.
Here’s an example of what's missing in the UI.
GA4 does great default automatic tracking of file downloads, videos viewed, etc. However, I believe it is impossible from within the UI to see what files were downloaded, or videos viewed.
With pageviews (1), you can see the page titles or page path (2):
For downloaded files, search terms or videos, however, you don’t get a similar table (other than in real-time, for the last 30 minutes). That file or video name is tracked in the relevant parameters, but the only way to see it is to build your own report to show it. (At least that’s been my experience so far – if somebody knows differently, I’m all ears!) I don't care how simple you want the UI to be - people will want to know which files got downloaded or which videos were viewed.
So why is it good that “less is more”?
Within the UI, I think Google expects the real insights to come from the tool itself, from its ML-driven (machine learning) insights. I have high hopes for this. I think this is one area where GA 4 may really shine and will probably get better over time.
But not all ML-driven insights are earth-shaking and, at the risk of being accused of a low blow, I don’t believe anybody is going to learn much from an insight like this one:
Less is Less – When the Data is Junk
In the new UI, some of the visitor’s demographic info appears to be hugely important, given the prominence it’s given in many of the reports:
But if GA 4 is going to display gender so dominantly, wouldn’t it be worth noting that that specific demographic is rarely captured for anywhere close to 100% of traffic? For example, here it’s captured for barely 25% of users:
Displaying that gender information so prominently, without at least an asterisk to warn the casual user that it’s not accurate at all, is misleading.
Pity the webmaster who is brand new to the field and who is setting up her first GA property today. She heeds the warning and installs GA 4. And now she has questions about setup, or interpreting data or creating her own reports.
Where on earth does she go for training? A year from now, there’ll probably be many decent training options; right now, it’s pretty slim. Charles Farina and Krista Seiden have good information available, for example, but it's still limited.
But if our new webmaster had installed Universal Analytics instead – holy cow, there’d be no problems finding training, reading blogs, watching videos, getting help from a jillion people.
Identifying Users (Anonymously)
GA 4 will identify users by a Google ID, the way they always have, and by Google Signals. Google Signals is a big step forward in cross-platform user identification – e.g., if the user came to your website on a desktop, tablet or mobile device, using Google Signals, there’s a good chance GA 4 will be able to identify that user as one person.
Google Signal relies on the user having a Google account and having opted into ads personalization. Given the huge popularity of Chrome, and the default setting of “on” for ads personalization, this will identify many users.
But user privacy is very important also – hugely important. Hence Google Signals only comes into play for websites that have “a monthly average 500 users observed per day per property."
That rules out quite a few websites.
If your website does a ton of e-commerce, then you’re in luck. Improving e-commerce appears to be a huge push within GA 4. The new e-commerce model is I think cleaner and richer than the model in Universal Analytics. The ML behind identifying user churn and users who are about to make a purchase – two of the most prominent new features in GA 4 – could be awesome.
So yes – GA 4 is potentially great for e-commerce sites.
But there are a ton of large websites that don’t have an e-commerce component, and which won’t necessarily benefit from a better e-commerce data model.
This turned into a bit more of a rant than I expected – all as a reaction to Google’s warning that Universal Analytics is not recommended for new Analytics users. Obviously, I disagree.
GA 4 is so radically different that if you are really going to start using it 100%, I think you should at least consider alternatives, like Adobe Analytics.
Luckily, it’s not an either/or proposition. You can run Universal Analytics AND GA 4 – which I recommend. Assuming you have an older version of GA installed on your website now, and you’re using it and getting good data out of it, then definitely install GA 4 as well. Start learning the interface; start taking advantage, perhaps, of some of its ML insights; start learning an event-driven analytics package. All good stuff.
In particular, roll out GA4 as a secondary analytics implementation if:
- you have apps you want to track
- your site does a lot of ecommerce
- you have more than 500 users a day, on average, visiting your site
But for heaven’s sake don’t throw out your old version of GA just yet!
Looking for more GA 4 tips? Check out our GA 4 Resource Hub!