Marketing attribution is a very useful tool for understanding how your users interact with different channels before coming to your website. This is very useful for valuing your marketing channels and understanding your customers’ behaviors. This tutorial will explain how to track marketing attribution in Google Analytics, specifically looking at multi-touch attribution.
What is Attribution?
Each time a user visits your site, they come from somewhere - by clicking on a search result link or clicking through an ad or even directly typing the URL to your site. These are known as marketing channels. Knowing where your traffic comes from, and where the valuable traffic comes from, is key to valuing your marketing strategy and to understanding your users. You can learn more about marketing attribution.
What is Multi-Touch Attribution?
Multi-touch attribution is best explained through an example. Let’s say Laila searches for sweaters on Google and clicks through a search result for sweaters.com. She browses for a while but doesn’t immediately purchase. She continues searching and browsing online. On a lifestyle blog about choosing the best sweater, they also mention sweaters.com and she clicks through. She’s still not ready to purchase, but she signs up for the newsletter. The next day sweaters.com sends her an email that she clicks through - and this time she makes a purchase.
This example involves three channels that bring Laila to sweaters.come before she makes a purchase: organic search (Google search), referral (the lifestyle blog) and email.
The most common attribution models in industry are single-touch models:
last-touch attribution, the default in most tracking programs including Google Analytics, gives all credit to the last channel before the conversion (email in our sweaters.com example) with the rationale that it’s the channel that closed the deal with the customer
first-touch attribution gives all credit to the first channel, with the rationale that i’s the channel that first introduced the brand to the customer - no future channels may have come into play without that introduction.
It’s easy to see the limitation in these models: what if the first and last channels (and the ones in between) were necessary for the conversion? This is where multi-touch attribution comes into play. It allows more than one channel involved in the conversion to get credit.
Google Analytics allows you to choose from the three most common multi-touch models. They differ in how they distribute the credit (weight) among the channels.
Linear - where each channel gets equal weight (in our example, 33.3% to organic search, 33.3% to referral, 33.3% to email)
Position based - where the first and last channel get 40% of the credit each, and the touches in between split the remaining 20% (in our example, 40% to organic search, 40% to referral, 20% to email)
Time decay - this is the most sophisticated model which assigns a larger weight the more recent a channel is to the conversion. So the last channel gets the most credit, the second-to-last channel gets a bit less (how much depends on how much time passed), and so on.
More details on the models here.
In our example, you can make the argument that Laila wouldn’t have clicked through the blog if she hadn’t already seen and visited sweaters.com right before when executing the Google search. Seeing the brand twice in such a short time frame could be a strong reinforcement. But what if that Google search had happened weeks before she saw the blog? Or months before? We wouldn’t feel as confident that Leila even remembered that first visit to our site.
This is where the concept of a lookback window comes into play. So how far back should we go when looking at the conversion path? Unfortunately, there is no globally correct answer. Google Analytics uses 30 days as a default and allows configuration between one and 90 days. If your product is one with a short purchasing cycle (like a sweater) you’ll want a shorter window. But if your product is one where people take a long time to decide (say, real estate or an expensive SaaS product), you’ll want a longer window.
How to Track Multi-Touch Attribution in Google Analytics?
Before you can track any type of conversions in Google Analytics, you have to have goals defined. You can refer here for more information on setting goals. In our sweaters.com example, we’re using the purchase as a goal. But we could have set many goals - signing up for the newsletter could be a goal, viewing multiple sweaters could be another, and so on. Conversion channels for each goal can be tracked independently.
Note that in the example videos we are using the Google demo account which is of a Google property. If you want to follow along and have the data look the same, sign up for the demo account here and in the date range in the reports select 1/1/2018 - 2/10/2018.
How to Track Multi-Touch Attribution in Google Analytics
- In Google Analytics, navigate to Conversions - Attribution - Model Comparison Tool.
- In the conversion dropdown, select the goal that you’re interested in measuring. In this example we’ll look at Goal 1: Purchase Completed and we’ll leave the lookback window at 30 days.
- Right below you’ll notice that Last Interaction (last touch) model is selected by default. Add in to compare to First Interaction (first touch) model and scroll down to the channel breakout.
- We see that when looking at the last-touch model, Direct accounts for over half of all conversions (1,660 conversions) with Referral accounting for roughly one third (973 conversions). But when looking at the first touch, we see that Referral greatly increases in significance. Now add in the position based model to give credit to all the channels before the conversion.
In this way you can easily compare the different models or select just the multi-touch model you want to use. You can click into the individual channels to see the channel breakouts with the model you’ve selected.
Marketing attribution allows you to understand user behavior and the value of your marketing channels, but you have to understand the different attribution models to get the most out of your tracking.