GA4 vs Universal Analytics for Ecommerce: the Good, the Bad and the downright Ugly!

Metageni works with a lot of forward thinking e-commerce businesses that have already transitioned to GA4 before the deadline of the end of June 2023 for Google’s sunset of Universal Analytics. We’ve been working through the knotty tasks of transitioning our customers historical raw multi touch attribution data across via BigQuery – Google Analytics raw database. We have also spent time reviewing ‘before and after’ report views in the front end, As GA4 superusers: here’s our early  thoughts:

The Good

The user interface is cleaner in GA4 than Universal Analytics and we can see how it will work better for the majority of more basic analytics users once some of the glitches have been fixed. 

And we totally agree that Google has done the right things for the underlying data, particularly the new and simpler event based logic. A ‘session’ or visit has always been a problematic metric due to the fact there is no clear way to define the session end apart from an arbitrary session timeout – so instead in GA4, we can view ‘session start’ events. This is a more straightforward and honest reflection of the data. Similarly transactions are based on ‘purchase events’. It all makes sense – although it is a bit strange to review the engagement event reports and see purchase events way down on the list under various far less relevant click related events like ‘view item’ and ‘apply filter’ and so on.

Another thing we really like about GA4 is the switch from bounced visits to engaged visits. The ‘bounce rate’ is a simple to understand measure – visitors who just leave without doing anything – but it is also a profoundly negative measure, leaving brands unsure what to do to remedy a high bounce rate. Far better to focus on positively engaging visitors and making those numbers bigger, than on the inverse problem of bounces. You still have the data – an engaged visit is a visit that does not bounce – but the metric frames the challenge in a more positive way that makes it clearer what is being measured and therefore what can be done about it.    

Last but not least, the ‘Explore’ custom report building is a great new addition and way better than the UA ‘customisation’ features. The set up has a Tableau light feel about it, which will put off less analytical minded and experienced users, but does mean that a much wider range of reports can be built. This is perhaps why the good folks at Google have focussed less on a report set of simple out of the box analyses, and more on the underlying data and a strong custom report interface.

The Bad

1. GA4 will only provide ‘Last click’ and ‘Data driven Attribution’ views with Google killing first click, linear, time decay, and position-based attribution views.  

Maybe they weren’t very popular reports and viewing these reports was a niche activity for Metageni’s superusers but why revert back to just Lack Click and DDA in the GA4 version?

Last click is just wrong for so many reasons we outline in this post and Google’s DDA is a black box and needs some rigorous testing vs other attribution methodologies. It is better than last click but it doesn’t include impression data and we’ve found this is the only way to truly understand the impact of massive digital marketing channels such as Instagram, TikTok and Facebook. We outline how we do that in this post here.. In one recent customer example we found 25 times more sales attributed to social platforms through impression tracking than last click alone : enough to justify continued spend.

It’s a hard UI to love: Universal analytics users were used to having a one screen dashboard of the key metrics: 

• Users
• Sessions
• Transactions
• Revenues
• Bounce Rates

GA4 hides this key information in the engagement event report and chooses to present all metrics from the top down based on the number of “counts” which drops ‘purchase event’ details all the way down to around Row 33 on one of the reports we viewed!  Definitely feels like Google’s engineering team were left in charge of key UI decisions without consulting performance marketing users.

The Ugly

• It’s a bit buggy and glitchy. Simple things like changing the date range of reports proves not to be sticky on scroll in the top user bar. The data changes but the date range reverts back to the original dates. That could do with a clean up.
• Unlike Universal Analytics: GA4 is reporting users by channel but not total users, sessions by channel but not total sessions. Why no column totals? Frustrating!
• What is the lifetime value chart in GA4? What does it even mean? More calculations and explanations would be helpful. 
• Much fewer reports: where is the great multi-channel funnels chart? Bring back the behaviour flow chart! These may have been niche activities for many but Metageni’s data scientists found them incredibly useful for digging deep into the data. 

Our summary: too much ugly and bad to outweigh the good at the moment but we hope Google are taking on board user feedback before sunsetting Universal Analytics for good and fixing the glitches and getting some of those nice superuser reports back. 

What it will mean is that users of GA4 are now more dependent on having the analytical skills to build their own reports. We can expect the ‘template gallery’ of custom reports to play a key role in bridging this gap – and would advise users to head to ‘Explore’ and play around with those if they are struggling to find their favourite reports in the new interface. 

Our biggest disappointment is the sunsetting of a lot of useful attribution reports in GA4 which increases the need for a custom attribution approach. 

Do you need help to get to reliable marketing measurement and ROI optimization?

If so just contact us: [email protected]