Google Analytics 4 with Cookieless tracking

How cookies messed up 52% of the Google Analytics 4 user data

Cookies are vulnerable - it’s not any news today but how much can they affect the reporting data, that's a completely different topic.

 

When server-side tracking started gaining momentum in the digital marketing community, it seemed like a holy grail - Solution to all the challenges that we faced back then. And it was somewhat true… Somewhat…

 

Standard server-side tracking systems, such as server-side GTM act as a data gateway between source and destination (Like GA4).

 

Yes, server-side tracking can bypass the ad blockers and tracking prevention on a certain level, that’s true but in this process the data quality somehow is left behind.

 

You might have heard about first-party and HTTP cookies as an alternative to solve the data quality issues. But today, after the latest iOS Safari and ITP update, they are nothing more than the plain old vulnerable cookies (Actually, first-party cookies are a more general term that combines both HTTP and Javascript cookies).

False feeling of safety with server-side tracking

In the first several months of server-side age, there were always talks about the increased number of purchases in GA4 reports.

 

And we also had the same cheerful talks on our first server-side tracking test when GA4 reported 96% of the purchases.

 

But there was already a feeling that something was missing.

What about the direct traffic?
How long does the user need on average to convert?
Do new users convert easily or do they need to get used to the brand first?
How first time users interacted with the website?
Are there recurring customers?
What else are we missing?

There are cases when the patterns of new users look similar to the recurring users’ patterns. And for a specific group of new users, they are actually identical.

 

What about the advertising?

What are the most common groups of ads that users interact with over time?
On average, how many ads do they interact with until the conversion?
Which was the initial ad that grabbed the user’s attention?

And why does GA4 report single ad as being sole contributor to the conversions? Is this true?

 

It wasn’t really possible to check the accuracy back then - We hadn’t even started thinking about cookieless tracking yet.

 

But today, we already have the EGO cookieless tracking system and these kinds of tests are just another Tuesday for us.

Cookies hit the data quality and your GA4 reports won’t be the exception as well

This article is not about third-party cookies or that they are being phased out. We solely focus on first-party cookies and in this case, specifically _ga cookie, as known as GA Client ID.

GA Client ID is an auto-generated unique identifier that Google tracking script generates and stores inside the user’s browser.

According to Google:
"'_ga', the main cookie used by Google Analytics, enables the service to distinguish one visitor from another and lasts for 2 years."

In most cases GA4 user data accuracy relies on this piece of data that is just 26 characters long, stored in the browser storage and can be deleted any time (Maximum in 7 days on iOS devices, so if your users mostly use iOS devices, keep that in mind).

We decided to measure how the cookies could affect Google analytics 4 reporting quality.

Comparing cookie-based GA4 VS. GA4 with COOKIELESS tracking

The objective of the test is to compare the data quality of two GA4 properties, one with the standard browser-side tracking and another with EGO cookieless tracking where the GA client id and Google Analytics itself is contained and stored on the server and powerful databases, never exposing them to the user’s browser.

 

Our goal was not to measure the data quantity improvements as any server-side tracking will increase the number of received events.

 

The website, where the test was gonna take place, already had an existing GA4 property. Even though it might have resulted in ambiguous results (And it partially actually did that - new users measurement, so we won’t cover them), we still proceeded and activated the new GA4 property linked to our EGO cookieless tracking.

 

The results were more concerning than we expected - at least 52% of the users' cases ended with split user data.

Data quantity improvement (Side-objective)

Even though data quantity measurement was not the goal of this test, we were constantly comparing real-time results - Just because we could.
Cookie-based GA4 property averaged 1300 concurrently active users.
while GA4 cookieless tracking had 15% increase to 1500.
We had similar results with purchase tracking:
EGO Cookieless tracking processed 100% of the purchases created on the store.
While the cookie based browser-side tracking ended up tracking just 74%.
We have to admit, while collecting events from the users' browsers, we were also receiving, combining and deduplicating data directly from the store back end that was processed and sent to GA4 immediately.

100% purchase tracking in this case was not something special but actually our system doing what it was designed to do - Combining multiple data sources with the historical data to achieve highest data quality.

What about the quality of the data in your GA4 reports?

52% USER DATA SPLIT

GA4 with cookie-based tracking

Cookie based GA4 tracking:

Falsely tracked at least 52%, total 86 000 users as two separate ones.
Altered user journey and sessions information where first part of the user journey was attributed to one GA4 client id while the rest of the journey was grouped under another one
User data split resulted in reduced average user engagement time - 3 times less than reported by GA4 with EGO tracking
52% of the user marketing data, including GA client id, was lost. (This can critically hit the advertising performance)
GA4 with Server-side cookieless tracking

GA4 with EGO Cookieless tracking

Prevented user data split for at least 52% of the users
Improved average user engagement time more than 3 times
Preserved marketing data, including GA Client ID, and managed consistent user tracking

How does the user data split affects marketing processes

To explain it simply, we can use an example - Let’s say the converted user had a total of 4 sessions on the website over the period of 2 weeks:
First one from Google Ads
Second one from Meta Ads
Third one from Google organic
And the fourth one was direct traffic where the conversion was made
Cookie based GA4 would show you that instead of one user with four sessions, there were two users, each with two sessions.

It would report to you that the second user made a conversion and while checking the channel conversion attribution you would find that only Google organic and direct traffic have 0.5 and 0.5 conversions attributed.

In the end, you might have assumed that neither Meta ads nor Google Ads had to do anything with the conversion and disabled them.

This is the issue with the cookies and analytical data quality - Data that is fragmented and leads to incorrect conclusions.

Conclusion

Cookies are vulnerable, you already know this. But we wanted to show you what this actually means for your day-to-day life. We wanted to emphasize the severity of this issue for all digital businesses and most importantly, to let you know that there are solutions.

But it is not the problem with GA4 itself. GA4 analytical system processes only the data it receives.

The goal of this test was to demonstrate how different tracking methods change data quality in GA4 and most importantly, how cookieless tracking can actually improve every piece of information you see in your reports.

First-party cookies are great until they aren’t and why should you only rely on them when you could get much more from your data.
EGO Digital Logo

EGO Digital