Understanding Attribution in Google Analytics 4 (GA4) vs Distil Analytics

Does GA4’s attribution model leave you with more attribution questions than answers?

It seems most marketers are in this position to some extent. After all, the road to migrate from Universal Analytics (UA) to Google Analytics 4 (GA4) has been a bumpy one filled with blind bends and black ice.

It’s been a journey of rethinking measurement objectives, deciding on an approach for year-on-year comparisons and considering whether analytics platform choice makes a difference. The short answer is it does – especially if you have clear expectations for your attribution insights.

In this blog we’ll explore:

Understanding attribution as a concept

What exactly does attribution mean?

Attribution is the process of identifying an outcome or situation as caused by a person or thing.

In the case of marketing attribution, it means identifying the source and value of the target outcomes based on which marketing activities created them.

Understanding marketing attribution is the foundation for return on investment analysis (ROI) or return on ad spend (ROAS).

What is a marketing attribution model?

A marketing attribution model refers to the method by which the impact and effectiveness of individual marketing channels is assessed. Calling it a model simply implies that there is a consistent approach used.

In any marketing attribution model, the touchpoints and conversions (outcomes) are assessed by individual asset, campaign and / or channel to understand how well they delivered the desired outcomes.

There are several different models to choose from when attributing success to specific marketing activities.

Most analytics platforms have their own attribution model. For instance, the Distil Marketing Analytics attribution model is a 30 or 90 day linear model. This means we capture touchpoints in the order they occurred in either the last 30 or 90 days and divide the conversion value equally by the number of touchpoints.

We think it’s the most equitable and inclusive way to assess marketing performance.

But other platforms, like Google Analytics in GA4, use a data-driven attribution model. In Google’s case, it uses data about every conversion and each touchpoint to decide how much value to attribute to each marketing interaction.

Why is accurate marketing attribution important?

It’s the basis of all marketing optimisation work and often budget decision-making. Marketers and business owners alike need a clear picture of how new and repeat customers make buying decisions, and therefore which marketing activities to invest in.

It’s important that marketing attribution is as granular and accurate as possible because it’s the foundation for understanding which marketing activities are driving which conversions, and therefore how best to spend marketing time and effort.

Understanding GA4 attribution & measurement concepts

What is Direct Traffic in GA4?

You might think ‘direct’ traffic includes only those visits that are from a bookmark or a typed URL into a browser – but there are a lot more users and sessions grouped in Direct than just those.

Direct traffic also includes any visits where there is no source information identified.

In GA4, Direct visits are excluded from receiving any attribution credit unless the entire journey consists only of Direct visits.

What is Unassigned Traffic in GA4?

Traffic is grouped into ‘unassigned’ when Google receives source information about the visit, but the source doesn’t match any of the other default channel groupings. This means there probably are UTMs included in the referring link, but Google doesn’t know how to classify them.

This is often resolved by moving to a standardised system of UTMs that match the default channels groups, since it’s not possible in GA4 to edit these groups.

However, Distil Analytics handles Direct and Unknown traffic differently. We don’t group Direct and Unknown – we separate them out. If a conversion has an identifiable session, we’ll class it as Direct. If there’s no session that can be matched to a conversion then it’s Unknown. This means you can have even more granularity in your marketing attribution.

Why is Unassigned Traffic in GA4 a problem?

Unassigned traffic presents a reporting problem because the group contains users and sessions that have visited your digital properties from any number of different channels. There could be email visitors mixed with referral visits mixed with organic social and other combinations of channels.

That means any ROI and ROAS assessments will be missing the contribution of each session and user currently grouped in Unassigned.

Imagine you had a particularly successful paid social campaign, but one of the ads was manually tagged and Google couldn’t identify the traffic coming from it as part of the same activity as users from the other ads. The visitors referred by this ad would be excluded from the success attributed to the campaign and therefore any return on investment calculations would be missing results.

Distil analytics works differently. We don’t group Direct and Unknown traffic – we separate them out. If a conversion has an identifiable session, we’ll class it as Direct. If there’s no session that can be matched to a conversion then it’s unknown.

How does GA4 attribution work?

GA4 offers three different attribution models in the platform.

1. Data-driven attribution

This is the GA4 default attribution model. Google defines it as “Data-driven attribution distributes credit for the conversion based on data for each conversion event. It’s different from the other models because it uses your account’s data to calculate the actual contribution of each click interaction.

The advantages of this model are that it will automatically calculate the value driven by each channel group and individual source. Without any manual analysis, you’ll get an ROI figure for each line. It requires no further setup when installing GA4, although we do recommend extending the default event and user data retention setting of 2 months to the maximum 14 months in order to give the model as much data as possible.

The disadvantage of this model is that there is no user control over how values are assigned to each touchpoint. The model is algorithmic in nature, specific to each account and its conversion events – and therefore different from the previous rule-based models.

2. Paid & Organic Channels – Last Click

This model ignores all direct traffic and attributes 100% of the conversion value to the last paid or organic channel that a customer engaged with prior to their conversion.

You might use this model if you want to see your ROI / ROAS figures assigned purely to the marketing channels where you have budget assigned. Ecommerce marketers sometimes find this helpful to have an at-a-glance view of purchase value delivered by channel.

The disadvantage of this model is that it discards the impact of brand marketing or other large-scale awareness campaigns that may drive direct traffic.

3. Google Paid Channels – Last Click

This model also ignores direct traffic and attributes 100% of the conversion value to the last Google paid ads channel the customer engaged with prior to conversion.

The advantage of selecting this model is that if your marketing budget is primarily assigned to Google Ads in areas like paid search, display and YouTube, this model will focus entirely on showing ROAS for that specific ad spend. In the instances where there are conversions using zero Google paid touchpoints, it will revert to the model above automatically.

Obviously, the limitation of this model is that it is entirely geared towards Google’s ad revenue. If that’s your primary spend area, then this may be a useful way to measure success, but for most users, this is not the best attribution to use. In the conversion paths where are user engages with other areas of marketing focus, such as email, organic social or organic search, these channels are given zero credit for assisting with the conversion.

For most GA4 users, the data-driven attribution model is the best option to choose, but it still leaves a lot of questions to be answered – which we talk more about in this section.

You can find the different GA4 attribution model options in Admin < Data display < Attribution settings.

What are the challenges of GA4 attribution?

The number of attribution models in the platform have been vastly reduced compared with Universal Analytics. This range has been reduced from the 6 different models that were available in Universal Analytics, and some may argue GA4 is the poorer for it. In my opinion, reducing the range of attribution models available limits a user’s capability to interpret the data in a way that makes sense for their business.

Google’s data-driven model offers no control over how different touchpoints are rewarded for their contribution to the eventual conversion. This means that Google is going to use whatever information it has about every conversion to decide, for itself, what the return on investment or ROAS is for each marketing channel.

The default reports in GA4 offer a limited view of attribution results. The User Acquisition Report shows only the very first channel users interacted with over time. Similarly with the Session Acquisition Report, this separates sessions from users, so it gives a clearer picture of the volume of activity, but again, only displays the source of individual sessions.

In my opinion, the range of default attribution reports in Universal Analytics (UA) was more comprehensive and useful. For instance, the Multi-Channel Funnel report in UA was extremely useful for looking at the role of channel groups in multi-touchpoint user journeys. It’s possible to build something similar in the Explore area of GA4, or if you have the Conversion Paths report in the Advertising section does a similar job.

There have been several reports of data discrepancies between GA4 and other internal tracking platforms. Volumes of events not matching up, the number of users differing and so on. This can be incredibly frustrating when trying to do accurate attribution and ROI calculations. However, these issues are usually not the result of an error in GA4 – although the data sets can take longer to process than those in UA.

Data discrepancies usually occur because of differences in measurement models, events triggering based on different conditions etc. Frustrating though this can be, significant discrepancies can usually be overcome by auditing both events, their triggers, conditions and values. For slight differences, you may have to adjust your expectations that data across two platforms might not ever match. For example, if a user rejects cookies in your website, Google Analytics won’t be able to place a cookie on their browser, yet your ecommerce platform will still record their visit and purchase.

How attribution works in Distil Analytics

As we’ve already discussed, every analytics platform has different attribution model options. GA4’s data-driven option offers little insight or flexibility into how value is assigned to the various touchpoints. In Distil Analytics, we’ve taken the opposite approach.

Linear Attribution Model

The attribution model in Distil is a linear model. This means that from start to finish, every touchpoint in the user journey is given an equal portion of the total conversion value. So if there are 5 steps in the journey, the value is split 5 ways.

Our linear attribution also has the option of a 30 or 90 day windows. This means you can decide whether you want to capture touchpoints in the last 30 days or 90 days. There’s no wrong choice here, it entirely depends on your business and how long your purchase cycle is. And you can easily switch between both models from our attribution dashboards to see what differences the time window makes.

And if you want to delve into the journey of each individual order – you can. Search by any attribute – order ID, customer email address etc or browse within a date range to see how you customer orders are placed by touchpoint.

Final Thoughts

The need for marketers to have a clear and robust understanding of the channels that deliver their goals in contrast with those that don’t has never been stronger. Customer data is more fragmented than ever, usually split over several different channels and platforms. Having the tech to overcome this is no longer optional if you intend to stay competitive.

Whether you’re new to marketing attribution or you’re struggling with the headaches created in the move from UA to GA4 – getting clear on your marketing attribution should be right up there on your priority to-do list.

FAQs

If you’re still stuck on GA4 attribution, here are some further FAQs that might help.

Q: What attribution model does GA4 use?
A: The default model is the data-driven attribution model. It uses an algorithm that’s specific to your account to assign value to all touchpoints (except Direct) in the user journey to conversion. You can read more about it in this section. There are two other options – one that covers paid and organic channels and one that focuses specifically on paid ads within Google’s network.

Q: How to change attribution model in GA4?
A: You can change the default model in GA4 by going into Admin < Data display < Attribution settings. You can see a screenshot in this section.

Q: What does data driven attribution mean?
A: For the purposes of GA4, data-driven attribution simply means the application of the GA4 algorithm to automatically calculate the value delivered by each marketing channel. You can read more about it in this section. In the broader context of the phrase, it means taking a data-led approach to understanding marketing performance.