Do you ever want to validate the source of the users on your website? It’s easy to track the source of your users through Google Analytics.
Google Analytics attribution modeling allows us to compare different models to figure out how they would impact our marketing attribution.
In this guide, we’ll understand and analyze the best attribution model for your website by comparing various attribution models on the Google Analytics platform.
An overview of what we’ll cover:
- Attribution concept
- GA reports by transaction
- Multi-channel funnels overview
- Model comparison tool
- Comparing attribution models
- Advanced attribution model
So let’s dive in!
Attribution Concept
You can validate the source of a user on your website by the process known as attribution. This process breaks down all the different sources that have brought the users to your website.
On your Google Analytics account, navigate to Acquisition → Overview.
You can also identify the high-performing sources that have brought you the most sales from the Conversions section.
However, it is important to understand the process by which Google Analytics decides the source which is attributed.
Attribution is a set of rules that give credit to the traffic sources for a particular conversion.
Let’s understand this concept with an example.
It takes a certain amount of time and a few interactions with various sources until a user becomes a customer.
As an example, let’s assume these interactions occur in a particular order.
The first interaction was when the user clicked on the Display ad for our website.
Following that event, the user also came to our website by searching through a Paid search.
Next, the user again visited our website by clicking on the Newsletter. But the conversion does not occur.
Finally, the user uses Organic searches to visit our website, and the conversion occurs.
The Google Analytics attribution method only gives the last source credit for attribution.
So, in this case, the entire credit for the attribution will go to the Organic sources, and the previous sources will be ignored.
Logically, this is not the best way to assign credit for the conversions. However, this is the way standard attribution models work in Google Analytics.
This method is called the last non-direct click.
As the name suggests, the credit for any attribution is given to the last non-direct source.
So, in this example, even if there was a direct click after the organic search, the credit will still be attributed to the organic source.
Therefore, it is essential to understand the reports that are affected by this method and whether you should spend time on attribution or not.
🚨 Note: Learning how to track UTM in Google Analytics can also help you improve your marketing skills.
GA Reports by Transaction
On your Google Analytics account, navigate to Acquisition → All Traffic → Source/Medium.
We’ll be accessing the E-commerce reports for the conversions.
Under the Transactions column, we can see the conversions made by each source.
According to our reports, google/organic had 3516 non-direct conversions. So it was considered the last traffic source in the customer journey path before the direct clicks.
This is probably not the best attribution method, but this is the default attribution method for Google Analytics.
To completely analyze the attribution, we need to understand the number of times each organic and direct source conversion overlapped with each other.
We’ll use Multi-Channel Funnels to analyze these overlaps.
Multi-Channel Funnels Overview
Conversions → Multi-Channel Funnels → Overview reports will show us the overlaps between the traffic channels.
By default, all the conversions are analyzed in the reports. Select the conversions you want to analyze.
In our case, we’ll choose Transaction conversations as we’re validating our attributions for conversions.
Click on Apply once done.
The overlaps between the particular channels will be seen in the Multi-Channel Conversion Visualiser.
For example, the Direct & Referral sources for conversions have a 25.58% overlap in our reports. Let’s name this Overlap A.
Additionally, we can also notice that the overlap between the Direct & Organic Search is 10.51%. We’ll name this Overlap B.
More importantly, you can also check the overlap between the Direct & Referral & Organic Search sources. This is 2.72% in our reports.
Let’s name it Overlap C.
If we want to find the total number of overlaps between all the channels combined, we can use the following basic math principle.
The total overlap will be Overlap A + Overlap B – Overlap C, as Overlap C is counted twice.
Hence, our overall overlap between the three channels will be around 33%.
We did these calculations for a total of three different traffic sources. But you can choose up to four different sources to generate reports.
This means that around 33% of our conversions are distorted by the last non-direct click attribution model.
As this model only gives credit to the last traffic source, it is viewed as an accurate report.
According to our experience, we have realized that if the overlap is less than 20%, then the reports won’t be distorted in most cases.
However, if the overlap is more than 20%, then we suggest you look at alternative attribution model options available in Google Analytics.
🚨 Note: Tracking funnels can be very useful for your marketing campaigns and you can do so with Google Analytics enhanced eCommerce tracking.
We can compare different models by using the Model Comparison Tool.
Let’s see how!
Model Comparison Tool
Navigate to Conversions → Multi-Channel Funnels → Model Comparison Tools.
This tool helps us to analyze different attribution models in Google Analytics.
Again, make sure you select only the conversions you want to analyze. In our case, these are Transaction conversions.
Now, we will show the alternative attribution models and how they attribute conversions and conversion values.
There are seven different attribution models by Google Analytics.
The first model is Last Interaction. As the name suggests, it only gives credit to the last channel, direct or non-direct.
The second model is the Last Non-direct Click. This is the default model in Google Analytics, and as we learned earlier, it attributes the conversion to the last non-direct sources.
The third model is the Last Google Ads Click. Again, as the name suggests, this model attributes the conversion to the last click of Google Ads.
This is especially useful when you’re optimizing your Google Ads for increasing conversions.
The fourth one is the First Interaction model. This model is exactly the opposite of the Last Interaction model, as it attributes the conversion to the first interacted source to the user in the conversion path.
All these four models give the credit for the conversion to one source only. However, the below three models are comparatively fairer.
The fifth one is the Linear model. This model evenly distributes the conversion and the conversion values among all the channels in its conversion path.
This means, that if there are two traffic sources, this model will give 50% credit to each source. Similarly, if there are four different traffic sources, this model gives 25% credit to each source.
Moving on, the sixth model is called Time Decay. This model is almost the same as the linear model, but it gives more credit to the sources that are towards the end of the conversion path.
Finally, the last attribution model is Position-Based. This model gives 40% credit each to the first and the last channels, and the rest of the 20% credit is distributed evenly among the other channels in the path.
So that’s how Google Analytics calculates conversions and conversion values using the different attribution models.
It is vital to understand the situations and scenarios for which any particular model fits the best.
Comparing Attribution Models
It is important to understand that there is no universal attribution model that works for every website.
Although the attributions are advanced analytics topics, we suggest you spend some time understanding them.
Using correct attribution models can be beneficial to you, as switching your marketing budget among traffic channels can be significantly high.
Switch the primary dimension to Source/Medium. This gives us a more granular view.
We’ll compare the Last Non-Direct Click, the default model in Google Analytics, to the Last Interaction model.
For our reports, we will choose to analyze Conversion & Value, and % change in Conversions.
The first traffic source is the direct source.
As we can see, it had about 7,000 conversions according to the Last Non-Direct Click model, and it had around 9,000 conversions for the Last Interaction model.
This suggests a rise of about 26% in the values.
Similarly, the second source is GooglePlex. This registered around 5,000 conversions for the Last Non-Direct Click model and about 4,000 conversions for the Last Interaction model.
The difference indicates a decrease of about 20%.
Let’s see how these models can affect business decisions.
As our goal is to bring new customers to the website, the Last Non-Direct Click model won’t work effectively.
Hence, we’ll use the First Interaction model in such cases.
Similarly, we’ll also use the channels that would close the conversion paths. Some of these can be retargeted or search ads.
In our case, we’ll filter our channel as google / cpc. Set the Secondary Dimension as Campaign.
Let’s compare our First Interaction model with the Last Interaction model.
You can see a significant percentage of change in the conversion distribution.
The changes range from -14% up to -50%.
As some of the conversions are at the very beginning of the conversion path, and some of them are near the very end of the conversion path, there is a significant difference in the values.
So, if you want to bring new customers to the website, you can tag the acquisition campaigns.
You can also compare the campaigns that are towards the end of the conversion path. Following this, you can choose the attribution model that fits the best.
You can use the First Interaction model to validate the acquisition campaigns, and the Last Interaction model to evaluate the campaigns at the ending of the conversion path.
💡 Top Tip: While comparing, never mix the acquisition conversions and the conversions towards the end of the funnel.
Next, you can compare the first interaction of the path and the last interaction of the conversion path. However, don’t compare all the interactions.
These are the basic attribution model comparisons that you can use to compare the first four attribution models.
You’ll need advanced attribution models to compare the last three attribution models, which distribute the values more evenly among the sources.
Advanced Attribution Model
Let’s compare the Linear model to the Last Non-Direct Click.
This comparison shows an increase in attribution to all the sources except the direct source. This is because we used the Last Non-Direct Click model for comparison.
On the other hand, if you compare the Linear and the Time Decay models, you won’t see much difference.
According to our reports, almost none of the sources have a difference higher than 10%.
This means that the results generated by the Linear and the Time Decay models are almost the same.
In the next step, let’s also compare a third model, the Position Based model with the Linear and Time Decay models.
You’ll notice that the difference in the conversion volumes isn’t significant.
So, if you want to use attribution models that don’t give credit to only one of the sources, you can use any of the three, either the Linear, Time Decay, or Position Based model.
You can compare all three models for your campaigns. If you find that the results are almost the same with any of the campaigns, we suggest you use those models.
Any change in your campaigns will also affect your business goals. Hence, we suggest you take sufficient time in understanding the models before finally switching to the most suited model for your needs.
🚨 Note: Since Google decided to sunset Universal Analytics 1 year from now, we recommend taking a look at our handy guide on how to upgrade to Google Analytics 4.
Summary
We can now see how to compare and choose the best attribution model that fits your business demands.
Attribution models will choose a particular source to credit for the given conversion by using the user data and conversion path they have followed.
However, if you want to manipulate how Google Analytics tracks the user path for conversions, then you need to create custom UTM parameters through Google Tag Manager.
How did the comparison values of attribution models differ for your campaigns? Which attribution model did you choose for your business? Let us know in the comments below!
