Multi-Channel Attribution Modelling

Users usually checked on different online channels through their conversion path. Marketing budgeting determined separately, independent from each other. However, they support each other in order to encourage users to create their conversions. 

For this post, I used Google Analytics dashboard to provide related information.

Conversion Paths

Conversion paths are simply user’s every interaction step until the path leads to a conversion/transaction. Google Analytics counts only interactions within the last 30 days, however you can change it up to 90 days. Conversion path data includes all online channels structured in Google Analytics: paid search, organic, direct, social media, affiliate, referral , email and so on.

Click on ‘Top Conversions Path’ under Conversions >> Multi-Channel Funnels section on GA.The Top Conversion Path report shows popular conversion paths and the channels (or touchpoints) a visitor interacted with before converting on your site. You can also search for specific channels/source/mediums. This report helps recognize if there is any channel sequence that tends to end with conversion.

What are multi channel attribution(MCA) models?

MCA models highlight the different perspectives on the touchpoints in the same conversion path until it ends with conversion. If you can track touchpoints properly and analyse models, you will be able to understand which channels help you to create conversion or which channels tend to be assisted channels.

There are different MCA models that you can apply on Google Analytics. Let me explain all of them through one example. Imagine a conversion path with 5 touch-points across different channels and ended with conversion revenue 50$. The sequence would be; Social Media – Direct – Paid Channel(Google Ads) – Affiliate – Direct. 

Last Interaction Model: 

All credit goes to the last step. So, if you use the last interaction model, MCA Dashboard would show this conversion brought by ‘direct’ channel and 50$ is only revenu from the ‘direct’ channel.

Last Non-Direct Interaction Model:

This model ignores direct traffic and attributes 100% of the 50$ credit to the last channel that the customer clicked through from before converting. ‘Affiliate’ channels would get 50$ for this case.

PS:Analytics uses this model by default when attributing conversion value in non-Multi-Channel Funnels reports.

Last Google Ads Click Model:

This model would attribute 50$ credit to the most recent Google Ads touchpoint before conversion. 

First Interaction Model:

50$ conversion credit goes first touch-step in the sequence: Social Media.

Linear Model:

Linear models distribute conversion value evenly across touchpoints in the sequence. In this case, every channel would have a 10$ conversion value. 

Time Decay Model:

Most of the credits would go to touchpoints nearest conversion. The Time Decay model has a default half-life of 7 days, the touchpoint interacted 7 days prior to conversion would take half of the credit, or  the touchpoint interected 14 days prior to conversion would take quarter of the credit.

Position Based(U-Shaped) Model:

This attribution model approaches attribution using weighted values. The most famous one is to assign 40% credit each to the first interaction and last interaction, and assign 20% credit to the interactions in the middle.

Extra: Data Driven MCA Model:

This model is only available in Google Analytics 360. Multi-Channel Funnels (MCF) Data-Driven Attribution allows you to create a custom model for assigning conversion credit to marketing touchpoints throughout the entire customer journey by using actual data from your Analytics account to generate. I am going to explain this future with an example scenario in another coming post. 

How to choose the right attribution model?

Since first and last interaction models ignore the whole conversion path, I would not recommend to consider them for the cases with conversion or CR KPI’s. If you create new campaigns across a couple of channels- with a minimal budget-  to create brand awareness, you may want to use a first interaction model in order to understand which channel is soaring traffic. Also, fist interaction is better for short- conversion cycles. Don’t forget GA counts interaction only up to 90 days. 

Last click attribution model appropriate for seasonal promotion campaigns when you have simple conversion paths. You can understand which channels tend to be direct converters. As your conversion path gets complicated, this model wouldn’t be an optimal solution. Touchpoints in the middle are important as well. Last non-direct click attribution model ignores last direct channels ended with conversion. It helps you understand which channel attracts your user to convert beforehand.

If you do not have an idea about which of your channels support your conversions or traffic consistently, a linear model would be a good starting point. Once you distribute equal credit across every touch point, you will observe their importance objectively. However, don’t forget if you have more than several same touch points in row, they will separately get their credit.Let’s say there are 3 direct touchpoints out of total 10 touch points on your conversion path, then, each direct touch point gets 10% of conversion value which equals to 30% in total.

Position based  is a good tool, if you care about both first and last interactions. You understand which channel is best for acquiring a new user and which one is excellent for conversions. This model is not a good option if you have consistently changing promotional campaigns. 

Lastly, time decay attribution model is better unless you focus on brand awareness. If you want to consider both assisted and last click touch point under the same segment but with different weights, time decay shapes weighting for you. Thus, you can give most of the credit the ones loser to conversion and less to ones in the beginning of the path.


I know it has been a very long post. However I want you to understand your options in MCA modelling better. You can use the Model Comparison Tool on GA in order to test different models’ effects on each channel quickly. Relying on only 1 channel causes neglecting the budget you spend for other channels. When you get confidence how to test these models, you are going to understand how systematically your channels work together. By the time, you may put different KPI’s and CPA’s for every channel. And, my friend, what we call for it is spending wisely.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s