GA4: The Countdown Is On

By Jaysen Gillespie, Head of Analytics and Data Science, RTB House

With the sunsetting of Universal Analytics (UA) just weeks away for free users of Google Analytics, it’s vital that marketers on the current system give some thought to what the shift could mean for their businesses. On 1 July, Google will force migrate a portion of the approximately 28.1 million websites using the platform towards Google Analytics 4 (GA4), and add another layer of complexity to the world marketers are currently having to navigate.

Marketers are already being asked to deliver even more with their marketing budgets, and they are under pressure to prove true return on investment (ROI), all while faced with the prospect of Google also removing third-party cookies from its Chrome browser in 2024. Such changes invite questions around how best to target audiences, and what to measure to demonstrate campaign effectiveness – two considerations that have been made even more important by Google’s decision to force people away from UA.

Getting ready ‘4’ the switch?

For marketers, there is a real risk that proving ROI in GA4 requires additional analytical effort, due to the complexity of the platform. That’s an issue that will particularly impact smaller companies that already tend to have fewer sophisticated, analytical resources at their disposal.  GA4 moves Google Analytics away from the comfortable pre-packaged reporting approach of UA (GA3) into a world most familiar to those of us in Data Analytics, with highly customizable “Explore” capabilities.

Both small and large companies should focus on accepting what’s not controllable and controlling what is controllable.

The new GA4 platform removes some lightly-used attribution models, such as “time decay”, while keeping the tried-and-true “last-click”.  GA4 also introduces a Google-calculated “data-driven attribution” option. While we take Google at its word that data-driven attribution is an unbiased attempt to assign credit fairly to all marketing vendors, we must admit that there is no way for marketers to validate the black-box approach in use.  We also think that the data-driven approach could potentially reshuffle credit in ways that are disruptive to current marketing plans.  As such, we advise marketers to stick to whatever their current attribution model is at this time.  Because GA4 will create enough changes for marketers to digest, we think it’s clearly in the interest of marketers to hold constant any settings directly under their control, such as attribution methodology.

Setting the GA4 purchase/conversion (event-scope) attribution model back to last click, if you’re already on last click, will mean one less thing to worry about. Once the platform has been studied, and there is an understanding of the potential benefits and challenges of the different attribution models, then it’s time to decide which attribution system actually makes the most sense for your business.  Even for marketers who retain the last-click setting, there are differences in how direct visits are handled and how sessions themselves are defined.  Keeping the attribution model constant allows marketers to digest the changes that will take place for all users..

The new attribution model at the forefront of GA4 is data-driven attribution, which uses machine learning to assess how different touchpoints impact conversions, distributing credit based on the likelihood of a touchpoint driving said conversions.

Data-driven attribution, while being a bit of a black box, also has great promise in some areas.  For example, if the algorithm lives up to its sales pitch, it should greatly decrease credit given to any fake clicks that may have previously been used to game the system. Attribution fraud, where dicey marketing services have their tag reload an ecommerce site to swipe attributed credit from a vendor sending a legitimate click, has been an issue in previous iterations of GA. Such fraud tricks the platform into believing a new session has begun, despite the click never actually occurring. However, the combination of data-driven attribution and a change to the way sessions are defined in GA4, should dramatically reduce the efficacy of this type of fraud.

Options, options, options

Though the shift to GA4 can’t be stopped, there are alternatives for marketers to consider.

First, it’s worth noting that those with the paid-for version of Google Analytics – Analytics 360 – will have an additional year before being forced to move over to GA4. So, there is an option for marketers on the free version to give themselves another year by signing up to the paid tier of the platform.

Alternatively, there’s the option of switching to a different analytics platform, or using the old school approach of web log analysers. In either of these cases, it would be important to carefully export as much data as possible from Universal Analytics.

Perhaps the most rigorous approach, however, is creating an in-house solution. This option provides full control and full transparency to media buyers and marketers, but it is mainly only open to the biggest and most successful digital-first businesses, due to the resources required to make it work.  One might also consider that GA4 itself, has elements of an ‘in-house’ solution built-in, such as seamless access to data in BigQuery and the ability to create “Explore” reports that are highly customizable.  Is it worth reinventing the wheel here?

The final countdown

As with any development, there are some early adopters, some in the middle of the pack, and then those laggards. It’s a bit late now to be considered an early adopter, but there’s still time to switch over to GA4 before Google forces everybody to.

Making business changes only when forced to is not a very practical way of running a business. It’s better to flip the switch manually than waiting for it to be flipped for you. The time for waiting is over – it’s now time to make a decision on what the shift to GA4 will mean for the future of your business. Are you ready?

GET IN TOUCH
Got a Question? We’ve Got Answers.