As the number of digital advertising channels increases, and consumers become savvier about online ads, it is harder to know which channels are the most lucrative. Attribution models are key to understanding which channels are performing, how effectively your advertising is working, and where the budget will be best spent.

Google introduced GA4 - its shiny, new analytics tool - to the world back in October 2020, and we officially saw the sunset of Google Universal Analytics on 1^st July 2023. Since then, the world has been wrapping its head around some of the major Google analytics changes. Our quick look at GA4 attribution models simplifies the changes and pinpoints its key features.

Attribution modelling is the process of assigning credit to a conversion from a click, an ad viewed, or another factor during a user’s path toward a conversion. As customers increase their time spent researching, comparing, considering and evaluating before making a purchase or an action of value, attribution modelling allows us to observe the path the customer took towards conversion, and pinpoint which touchpoint inspired a conversion.

This can also help to identify which channels may be in need of some adjustments and optimisations. For example, your biddable media accounts are performing well and achieving high click-through rates, but if your product pages are not optimised for SEO, you may miss out on valuable conversions. Attribution models can help identify these missing opportunities.

There are three categories of GA4 attribution modelling:

1. Data-driven attribution modelling
2. Cross-channel rules-based modelling
3. Ads-preferred last-click modelling

A data-driven attribution model in GA4 deeply analyses conversion path data based on your conversion events. The aim is to understand how different marketing touchpoints impact the user’s likelihood of converting.

It uses a combination of machine learning algorithms and collected data (where data availability allows).

There are two key elements in data-driven attribution modelling:

1. An analysis of collected conversion data to create conversion rate models.
2. Combining the conversion rate models with machine learning to assign conversion credit to each channel.

Data-driven attribution modelling also takes a number of factors into account in order to assign conversion credit to a channel. These are:

• The time between an interaction and a conversion
• Device type
• Number of ad interactions
• Type of creative asset
• The order of ad exposure

Cross-channel rules-based models assign value to advertising touchpoints based on a set of predefined rules.

• Paid and organic last-click modelling

This rule ignores direct traffic and assigns 100% of the conversion value to the last click (or engagement for YouTube) before a user converted.  Direct traffic refers to users converting directly on a platform without having clicked on other advertising touchpoints. Direct traffic will only gain conversion value if all the touchpoints leading to a conversion come through direct traffic.

Example:

1. Social > Display > Paid Search > Organic Search = 100% credit to Organic Search
2. Display > Organic Search > Direct = 100% credit to Organic Search
3. Organic Search > Paid Search > Direct > Email = 100% credit to Email

Note: Prior to May 2023, other rules including first click, position-based, linear and time decay were optional conversion rules-based models. However, Google deemed these as inflexible to the changing consumer journey, and retired these options for new GA4 users from May 2023.

This GA4 attribution modelling assigns 100% of the conversion value on the last Google Ads channel clicked through before converting, regardless of whether it was the last touchpoint before a conversion. If, however, no Google Ads are clicked before a conversion, conversion value is assigned to the last paid or organic click.

Example:

1. Social > Display > Organic Search > Paid Search = 100% credit to Paid Search
2. Display > Paid Search > Organic Search = 100% credit to Paid Search
3. Organic Search > Social > Email = 100% credit to Email (attribution model reverting back to last click modelling as no Google Ads were clicked)

Note: Conversion data can still be collected up to seven days after a conversion is recorded. Choosing a date range prior to these seven days will offer a more accurate analysis.

GA4 uses data-driven attribution modelling as its default attribution setting. Google prioritises this attribution modelling as its use of advanced AI machine learning is more capable of taking into account the evolving consumer journey and places an increased emphasis on cross-network conversions.

It’s not possible to say definitively which attribution model in GA4 is the best. It often depends on the priorities of the business, your chosen conversion events, and the paths your converters take.

Data-driven attribution model in GA4 goes a long way in offering an intuitive analysis of conversions, more so when rich data is already available. However, it’s important to take a holistic approach when analysing conversions.

This is where attribution model comparison is key. This feature within GA4 allows businesses to compare data collected through different attribution models, and assess which model yields the most accurate data. Comparing attribution model data can also help to uncover overlooked campaigns that are significant in aiding the conversion path.

GA4 attribution models can be key in understanding a consumer’s path towards a conversion, and is useful in aiding decisions on advertising channel prioritisation. However, it is important to take a holistic approach when making any major changes to advertising, and to never underestimate seemingly minor touchpoints within a buyer’s journey.

If you need any help or advice navigating GA4, or setting up a new account, contact our team of knowledgeable digital marketers today. Data underpins everything we do at Zelst, so we’re here to help you measure and understand your consumer’s conversion touchpoints. Take a look at the results we’ve achieved for our clients so far.