What Attribution Model Does Google Analytics Use?

Google Analytics is a staple part of most marketers’ toolkits. Along with Adobe Analytics, it makes up the majority of the analytics tool market.

But with the impending removal of third-party cookies in Chrome, continued questions about its legality and important changes to their newest iteration, GA4; Google has increasingly been put under the spotlight.

And as such, marketers are understandably considering their options for accurate, and compliant, marketing measurement.

So, what level of attribution is really available within Google Analytics – and what attribution models does it use to provide this?

Read on, as we dive into this topic:

• What attribution models are available in Google Analytics?
• How does Data Driven Attribution work?
• What attribution models are not available in GA4?
• Why is Google removing these models?
• The problem with using Google Analytics for attribution

What attribution models are available in Google Analytics?

Historically, Google has received a bad rap for it’s over reliance on Last Click to provide its users with attribution reports.

However, from September 2021, they moved away from this and instead introduced a new Data Driven Attribution (DDA) model as their default.

Marketers do have the option to opt-out from using this, as Google has kept five rules-based attribution models in the mix:

• Last click
• First click
• Linear
• Time decay
• Position-based

However, some of these are only available for a little while longer – as we’ll cover later in this blog.

How does Data Driven Attribution work?

Google’s DDA uses Machine Learning to understand how each marketing touchpoint contributed to a conversion, while remaining privacy compliant.

The advertised benefits of this new model include:

• Learn which keywords, ads, ad groups, and campaigns play the biggest role in helping you reach your business goals
• Optimise your bidding based on your specific account’s performance data
• Choose the right attribution model for your business, without guesswork

What attribution models are not available in GA4?

In April 2023, it was announced that Google would be removing some core attribution models from its platform, as well as Google Ads:

• First click
• Linear
• Time decay
• Position-based

In May 2023, these models will be unavailable for any new conversion actions.

From June 2023, Google will begin removing the ability to select these models for ads that don’t already use them for reporting within Google Ads.

They will then be officially sunset in Google Analytics and Google Ads beginning in September 2023.

Why is Google removing these models?

According to Google, these attribution models are being sunset due to “increasingly low adoption rates, with fewer than 3% of conversions in Google Ads using these models”.

Following the announcement from Google, Corvidae’s Senior Software Sales Executive, Pete Allcock, shared some thoughts on the news over on LinkedIn:

“Despite the vast majority of businesses still using Last Click as the preferred model; on the face of it, it seems a weird decision to remove features from a new platform which is still being developed.

“But, when you consider what we are actually dealing with; it all begins to make a little more sense…

“First party cookies really struggle to connect a customer journey cross-device/browser/user-agent etc. As a result, upper and middle funnel are disconnected.

“So, regardless which model you pick (this includes Data Driven Attribution – DDA) you will be skewed towards Last Click anyway. As a result, we are really looking at Last Click vs Almost-Last Click.

“With everyone using the same measurement protocols, the inevitable happens, and we all end up fishing in the same pond – largely at the bottom of the funnel; which further impacts CPA.

“So, who benefits from increased ambiguity and increased cost?! I heard you ask.

“Far be it from me to suggest, but surely, it’s the very platform you’re spending your budget with; the same one that provides the measurement you’re basing your decisions on.”

The problem with using Google Analytics for attribution

As Pete covered in his LinkedIn post, the real problem with marketing attribution doesn’t sit with the attribution model.

It comes down to the way that data is gathered and collated in the first place.

As such, cookies aren’t fit for purpose. They provide marketers with disjointed user paths that are impossible to optimise.

And while Google are soon to deprecate third-party cookies from Chrome and have moved to a Machine Learning-focused model, this doesn’t solve the underlying issue.

CEO and Founder, Chris Liversidge, summarised this in a recent webinar:

“GA4 & Adobe are heavily reliant on the cookie to stitch paths together.

“Use of Google Signals and ID 2.0 3rd party data sharing have little to no impact on stitching success.

“Cross-device user journeys are impossible to effectively track and optimise using GA or Adobe, and Data Driven Attribution (DDA) is siloed to those broken paths, creating misleading results.”

As we can see in the graphic above, by relying on cookies to track journeys, once a user switches to a different device, it splits a full journey into two separate paths.

If we were to apply Google’s DDA model to the converting path (Path B), we’re applying more revenue than we should to the three touchpoints that are visible because we’ve excluded the the touchpoints that were in Path A.

So, we’re left with overinflated reports and a skewed view of how our marketing activity is really performing.

Is there an alternative to Google Analytics?

As we mentioned at the start of this blog, GA is a core part of the marketing toolkit for most. So, considering a move away from what is familiar can seem daunting to say the least.

But there is a better way to do attribution.

We’ve spent almost a decade researching attribution methods and have developed a patented approach using AI to remove the need for cookies to stitch together complete user journeys.

Corvidae creates user paths that are on average twice as long as Google Analytics and operates at over 95% accuracy compared to just 20% for cookie-based methods (used by attribution tools like GA and Adobe).

Learn more about how our unique approach to attribution compares to Google Analytics. Or, to find out how you can get started with Corvidae, download our guide below.