Best Channel Structure and how to think about Attribution (in 2021)

  • What channel segmentation to use?
  • What is Attribution and why is it so tricky?
  • How to use Google Analytics Attribution Tool 
Image of co-founder

by Al Taylor
Updated June 28th 2021

Section 1:

What Channel Segmentation to use?

How does a web analytics tool such as Google Analytics Track Visitors to your site?

By using Tracking Code and a Pixel to send the information to the GA server.

Channels are a way of putting sessions into buckets that match your different and separate initiatives that drive visits to the site.

You can then work out how each initiative is working and what to do more or less of.

Channels are fundamental to being able to optimise your e-commerce store.

One key element of Channel Segmentation is to not mix paid and free visits in the same channel.

What’s the best channel segmentation?

There isn’t a definitive list of Channels.

Google Analytics now offers three channel segmentations:

The original ‘Default channel grouping’ from when Google Analytics was first launched (and which is now not fit for purpose)
A channel segmentation they use in their Attribution tool
A revised ‘GA4 Default channel grouping’ for their new GA4 web analytics tool

A comparison of the three channel segmentations is below:

Table comparing GA channels
We recommend the channel segmentation that is used in the GA Attribution Tool. This segmentation splits out Paid Social (which the GA Default channel segmentation does not) but also because it separates out Shopping and Paid Search Brand (both of which can behave very differently to Paid Search Generic and attract a different type of customer or awareness level of brand). This is a summary of what the different channels mean:
Table explaining what different channels mean

if you are running specific Paid Brand Campaigns (an ad from these Paid Brand campaigns should be triggered if the user uses one of your brand keywords – such as your company name or one of your own brand product names in a search) then make sure the word ‘Brand’ is in the campaign name so that costs from brand campaigns can be easily grouped and attributed.

We recommend that you do run Paid Brand Campaigns.

Users that are searching for your company tend to have much better conversion than the average user as they are already familiar with your brand and may well have purchased before – we don’t want to mix these people in with other paid visitors who have lower brand affinity.

It’s really important that any visits from your paid campaigns accurately capture the Campaign Name, Source, Medium (and Ad Set mapped to Content for Facebook campaigns) using UTM Parameters.

You can read how to set up UTM Parameters for Facebook and Google in this article.

You also need to make sure that your Campaigns are named sensibly to capture key information about them. You can read more about naming campaigns in this article.

If GA’s Default channel grouping is not right for you then mapflo can help re-assign sessions and conversions to your preferred channel segmentation.

mapflo can use data from your ad platforms to perfectly organises and segment you marketing spend data – providing more granularity around the type of paid campaigns driving sessions and purchases: category of product advertised, brand vs non-brand; purchase vs brand awareness vs lead gen).

Section 2:

What is attribution and why is it so tricky?

  • What is attribution?
  • Why conversion path / customer journeys make perfect attribution impossible?
  • GA default attribution: non-direct last click

It’s easy for Google Analytics to track information about each visit/session on your site and assign each visit to a channel.

What is more complicated is attributing a channel to a purchase made on your site.

The user who made the purchase may have visited your site multiple times in the preceding days, weeks and months before making the purchase and used multiple channels across those visits.

Because Google Analytics assigns a unique clientID to each visitor (read more about tracking) then it is possible to see the unique conversion paths / purchase journeys (sequence of visits by channel).

Here are some example conversion paths:

Example customer journeys

Say a user initially came to the site via Paid social, but on the visit that led to the purchase the user came through Organic search then which channel do you attribute that purchase to?

The situation is more complex still

The conversion path tracks users on the same device (as the clientID is based on the combination of browser and device).

In real life people may use multiple devices. Google Analytics will treat each device as a separate user.

Additionally, a person may see an ad, not click on it but go to your website directly or through another channel (this is called a view-through conversion by the ad platforms).

Google tries to address the first issue about multiple devices through it’s Google Signals product which can unify visits from the same user across multiple devices but only if that user is logged in to their account on your site and has the correct marketing permissions.

Facebook reports its own conversion data based on a view-through window – it can track if someone that has viewed your ad (but didn’t click on it) but then subsequently made a purchase within a set time frame (the window).

GA default attribution Model – Non-direct Last Click

The default attribution model in Google Analytics is ‘Non-direct Last Click’ (which means the click that led to the purchase gets all the credit unless the last click was Direct in which case the most recent non-direct click is given all the credit).

We think last non-direct click is a reasonable attribution model. For some ad platforms like Google Ads we think it overall reflects the contribution as well as any. For other ad platforms such as Facebook it is likely to underestimate the impact and you will need to perform some kind of test to understand how by how much it is underperforming so that you can give additional credit to Facebook.

You can use the Google Analytics Attribution Tool (see next section) to look at the different models and if you think the last non-direct click model under or over represents a particular channel then you can use mapflo to give certain channels a boost.

Section 3:

Attribution Tool in Google Analytics

If you use Facebook Ads then you should be setting up UTM Parameters on every Campaign – the good news is that it is super simple

image comparing A and B models

The Attribution Tool in Google Analytics is really helpful for kicking the tyres on attribution.

It shows the variation in conversions a particular channel gets credit for depending on the model used (and in our experience the variation is not that different between models)

It lets you see:

  • Conversions using the Attribution Channel Grouping (which includes shopping and Paid social)
  • Conversion paths (how many conversions by unique journey types)
  • Comparison of attribution models by channel (conversions attributed to each channel using different attribution models) – this is really useful in giving you a sense of which channels benefit/lose out depending on the model)
  • Comparison of attribution models by Google Ad Campaign – at a more granular level this can again give you some idea as to which campaigns benefit/are penalised by different attribution models

There are six different attribution models you can view and compare your conversions by:

  • Last click: Gives all credit for the conversion to the last-clicked event.
  • First click: Gives all credit for the conversion to the first-clicked event.
  • Linear: Distributes the credit for the conversion equally across all clicks on the path.
  • Time decay: Gives more credit to clicks that happened closer in time to the conversion. Credit is distributed using a 7-day half-life. In other words, a click 8 days before a conversion gets half as much credit as a click 1 day before a conversion.
  • Position-based: Gives 40% of credit to both the first- and last-clicked event, with the remaining 20% spread out across the other clicks on the path.
  • Data-driven attribution: Data-driven attribution distributes credit for the conversion based on observed data for each conversion type. It’s different from the other models because your account’s data is used to calculate the actual contribution of each click interaction. Each Data-driven model is specific to each advertiser and conversion type.

How to access the Google Analytics Attribution Tool

You can see the channel grouping for the attribution tool by clicking on Attribution (bottom left) in Google Analytics (you may have to activate).

You land on the ‘conversion paths’ page (this is the example Google Merchandise Store and I’ve filtered for conversion paths that include Paid search).

Landing screen attribution tool
Click on ‘Model Comparison’ on the left hand menu:
Screenshot showing Attribution Tool Google Merchandise Store - Model Comparison - default channels
You can see that it displays the default channel grouping (header in the first column of data) – by clicking on the edit button to open report settings in the right hand column you can then change the dimensions to Channe (click on Analytics default channel grouping):
Change to channel from default channel grouping on attribution tool

Now the model comparison is segmented using the attribution channels (rather than GA default channel grouping)

Screenshot showing Attribution Tool Google Merchandise Store - Model Comparison - channels

The Google Ads performance view is really helpful at providing some insight into which specific campaigns perform best under different attribution models. If you use non-direct last click as your standard model but can see that some campaigns have a significant uplift in conversions under first click then you might be happy to push those campaigns a bit harder.

Screenshot Attribution Tool Google Merchandise Store Google Ads performance

It’s easy for Google Analytics to track information about each visit/session on your site and assign each visit to a channel.

What is more complicated is attributing a channel to a purchase made on your site.

The user who made the purchase may have visited your site multiple times in the preceding days, weeks and months before making the purchase and used multiple channels across those visits.

Because Google Analytics assigns a unique clientID to each visitor (read more about tracking) then it is possible to see the unique conversion paths / purchase journeys (sequence of visits by channel).

Here are some example conversion paths:

Example customer journeys

Say a user initially came to the site via Paid social, but on the visit that led to the purchase the user came through Organic search then which channel do you attribute that purchase to?

The situation is more complex still

The conversion path tracks users on the same device (as the clientID is based on the combination of browser and device).

In real life people may use multiple devices. Google Analytics will treat each device as a separate user.

Additionally, a person may see an ad, not click on it but go to your website directly or through another channel (this is called a view-through conversion by the ad platforms).

Google tries to address the first issue about multiple devices through it’s Google Signals product which can unify visits from the same user across multiple devices but only if that user is logged in to their account on your site and has the correct marketing permissions.

Facebook reports its own conversion data based on a view-through window – it can track if someone that has viewed your ad (but didn’t click on it) but then subsequently made a purchase within a set time frame (the window).

GA default attribution Model – Non-direct Last Click

The default attribution model in Google Analytics is ‘Non-direct Last Click’ (which means the click that led to the purchase gets all the credit unless the last click was Direct in which case the most recent non-direct click is given all the credit).

We think last non-direct click is a reasonable attribution model. For some ad platforms like Google Ads we think it overall reflects the contribution as well as any. For other ad platforms such as Facebook it is likely to underestimate the impact and you will need to perform some kind of test to understand how by how much it is underperforming so that you can give additional credit to Facebook.

You can use the Google Analytics Attribution Tool (see next section) to look at the different models and if you think the last non-direct click model under or over represents a particular channel then you can use mapflo to give certain channels a boost.