From the inception of the ecommerce sector, businesses worldwide have worked tirelessly to develop their understanding of consumers’ online behaviour, pinpointing where, why and how customers are interacting with websites and online advertising. The primary solution to this seemingly gargantuan task? Cookies.
For those not in the know, a ‘cookie’ refers to a minute file containing a small piece of data, allowing advertisers rich insights into user journeys, with browser history, location, on-site actions and IP addresses among the plethora of trackable metrics.
This is, of course, old news. Not only has this method of data collection existed in one form or another since the mid 1990s, but has earned quite the reputation for controversy. Users have little control over the data they share, or the ways in which said data is used.
That said, concerns regarding the use of personal data have not gone unnoticed and digital landscapes are shifting. The introduction of tighter General Data Protection Regulations (GDPR) in the EU and EEA (European Economic Area) has set the path to a new future, one in which third-party cookies are no more and machine learning algorithms take centre stage.
In this article, I want to explore the power of machine learning within Google Analytics 4 and the role that such tech will play in a cookieless future.
What Does A Cookieless Future Look Like?
Before diving into the technicalities of a ‘cookieless’ digital future, it is worth clarifying the difference between first and third-party cookies. First-party cookies are snippets of data only accessible by a website’s owner. These play an important role in the collection of analytical data and optimisation of site functionality. Third-party cookies, on the other hand, are created by separate domains and can track a user’s behaviour across the internet.
Cookieless, in this instance, refers to the abolition of third-party tracking across internet service providers (ISPs). First party cookies are here to stay, for the time being.
It goes without saying that scrapping third-party cookies will leave significant gaps in the puzzle of online marketing. Google, in particular, relies on cookies to fuel its advertising platform. However, for the sake of privacy compliance, alternative methods must be utilised to fill in the blanks.
We’ll explore this in further detail below, but it is worth noting that tools such as statistical modelling, predictive analytics and machine learning will prove invaluable as we step into a cookieless future.
Machine Learning In GA4
As of July 2023, Google’s Universal Analytics software has been permanently retired with Google Analytics 4 (GA4) superseding it. To put the scale of this into some perspective, the tech giant’s current analytics property is used by more than 20 million websites worldwide and, with the transition to GA4 now complete, there are a number of pretty big changes digital marketers are faced with.
In addition to a sharpened focus on user privacy, GA4 places a keen emphasis on its machine learning capabilities. ‘Cookieless tracking’, as it is known, relies on ever-improving AI algorithms to connect the dots in lieu of third-party cookies. By utilising a blend of directly observed and modelled data, GA4 is able to build a comprehensive and accurate dataset, without breaching user privacy regulations.
What’s more, AI integration and machine learning has the potential to provide more accurate data than that collected by third party cookies. By combining predictive analytics, analytics intelligence and behaviour modelling, GA4 is able to compensate for data lost as a result of cookieless browsing. Below, we’ll explore the specifics of these features.
Predictive Modelling
Among GA4’s most innovative features is its ability to predict future metrics and audiences. From the moment a Google Analytics account commences data collection, the property’s built-in machine learning algorithms begin to learn from this information. By analysing account-specific datasets, GA4 is then able to generate projections such as revenue, events and trends.
Data gathered by GA4’s machine learning algorithm can then be used to create predictive audiences. In other words, based on past events, GA4 is able to forecast which of your audience is most likely to churn. This information may then be fed into Google ads for the purpose of churn prevention, such as remarketing campaigns.
Analytics Intelligence & Anomaly Detection
Moreover, GA4’s Analytics Intelligence will play an important role in the navigation of a cookieless future. In Google’s own words, Analytics Intelligence is ‘a set of features that uses machine learning and conditions you configure to help you understand and act on your data’. Not only does machine learning have the capacity to recognise and syphon anomalies and errors, it will change and adapt as a result of these inconsistencies. In other words, the more unique data GA4 receives, the better it becomes at detecting outlying information.
GA4’s ability to separate genuine data from anomalous results represents a positive step away from third-party cookie reliance. While cookies are believed to be up to 60% accurate at the best of times, machine learning and analytics intelligence presents the opportunity for perpetual improvement.
Behaviour and Conversion Modeling
Integral to GA4’s data compliant existence is the introduction of ‘consent mode’. Essentially, this enables businesses to automatically generate predictive data when a user does not consent to collection. Should an individual decline consent, GA4 draws from relevant pre-existing data to generate a prediction of the declining user’s data.
This form of behavioural modelling offers invaluable insights when directly observable data is unavailable. Rather than leaving cookie-shaped voids in your data sets, GA4 enables observed data to be blended with behaviour models, providing a far more complete overview of user journeys and site interactions.
What Does Machine Learning Mean For Marketing And Data Analytics?
What is particularly exciting about machine learning within GA4 is its accessibility. Whereas past examples of machine learning were only available via third-party solutions, Google’s integrated features place the power of AI at marketer’s fingertips. Of course, as is the case with all machine learning models, the algorithms require data to improve; the more unique input they receive, the better they perform.
The real question is, what does machine learning mean for digital marketers in a cookieless future?
First and foremost, it demonstrates the need for adaptability and technological open mindedness. Machine learning algorithms, such as those within GA4, are certain to play a major role in filling the gaps left behind by third-party cookies. While GA4 may not be perfect (yet), businesses would be foolish to overlook the potential of machine learning within the property.
Moreover, machine learning algorithms signify a change in the ways we observe and utilise user data. As we reduce our reliance on third-party cookies, features such as predictive modelling, analytics intelligence and anomaly detection will take up the gauntlet.
Paul Morris is the Managing Director of Superb Digital. He has over 20 years experience in the field and has written for the likes of Search Engine Watch, Business West, Business Leader and Search Engine People.
You can connect with him on LinkedIn here.
