Google Analytics to Fuel a Healthy Discussion About Attribution

Set the Stage – The High-level Metrics Most Will Stop At

One of the most common reports a marketing manager might look at is the Acqusition > Overview breakdown of each channel type (e.g. Organic, Paid Search, Email, Display, Social, Other, Referral, Direct). This gives a great summary of how many new users, sessions, transactions (or other goals), revenue, etc… that each channel is contributing. For most, this is as much as they feel they need to know to inform further investment decisions and evaluate the success of the channels they manage.

Google Analytics Acquisition Overview

Google Analytics Acquisition Overview

Unfortunatley, this is a last click view into a multi-influence world. What is one of the easiest ways to tell that this view needs a different perspective?


Direct traffic (where someone types your URL into the browser address bar and heads straight to your site) cannot be the first interaction with your business. Someone has to learn what that address is from somewhere. Even if they luck out by typing the name of your business into the browser and adding “.com”, they still had to learn the name of your business from somewhere. Direct doesn’t just happen.

Given this, and knowing that we want to be able to start discussing how we are influencing our own success, we need to start looking at the things that happen before the transaction (or other goal). Thus enters the MultiChannel Funnels > Top Conversion Paths report in Google Analytics.

Learning More About the Customer Journey

We all hear this phrase many times when it comes to marketing: “the customer journey”. The Multi-Channel Funnels > Top Conversion Paths report in Google Analytics will identify, by clicks to the site, the interactions that customers who have converted (purchased, signed up for a newsletter, etc…) have had with each channel.

It is quite common for the top conversion path to be “Direct”, “Organic”, or “Paid Search”, thought this will vary depending on your business’s marketing mix and outreach strategy. In the example below, “Direct” makes up the most common conversion path.

Google Analytics Top Conversion Paths

Google Analytics Top Conversion Paths

Remember, though,

Direct doesn’t just happen.

As early as the 6th and 7th most common conversion paths in this example, we can see that, at least based on clicks to the site, organic and paid search traffic lead to conversions that happened via direct as the last interaction. In a last click view, these would get rolled up under “Direct”, and we might assume that those were people who had visited the site maybe months and months ago (and thus are outside of our look back windows or cookie duration), when, in fact, they are likely new customers who found the site through efforts (and spent resources) put into SEO and SEM.

Pulling up this list of paths, and getting into a room with all of the heads of the marketing channels, can be extremely useful in fostering a conversation about how customers find you, gain trust in you, and, eventually, do business with you.

A useful follow up to this view, for me, has been to filter this report by each channel, once, so that the head of each department can see all of the ways that their channel has contributed to business goals. For example, below, I’ve filtered the report to show any path that includes “Display”. Note how incredibly complicated the journey already reveals itself to be, just on the third most common conversion path.

Google Analytics Top Conversion Paths Filtered For Display

Google Analytics Top Conversion Paths Filtered For Display

In this example, 437 out of 1,497 transactions included major influences by the display channel. According to last click models, this channel only resulted in 187 transactions. However, we can see in our first screenshot of this example, that this channel brought us a whopping 49% of our new visitors. That’s an incredible contribution, and now we can see how often that lead to an eventual transaction.

What Model Works … to Start With?

Once every channel has had a chance to review their contribution and places within the customer journey, and the entire marketing department can get together to talk it through, the natural follow-up question is going to be, “now what?” We want to start making our marketing investments where the influence helps, which means that we may not want to stick with a last click attribution model, but with such a complicated environment in which to work, what model do we choose to evaluate this?

Every situation is going to be unique, and, as long as people learn and departments get more efficient at what they do, the model will never be perfect. However, it is helpful to see, of the standard models available, how does that change the “return” that each channel is realizing for the business.

Here, we introduce the Attribution > Model Comparison Tool in Google Analytics. This report allows you to compare 7 predefined (and most common) models, three at a time, and even compare custom built models. The four that I see used most frequently are the “Last Interaction” (since it’s the default that most marketers are used to, and, therefore, makes for a good baseline comparison), “First Interaction” (which helps identify the value eventually driven by first making a potential customer aware of our business), “Time Decay” (which visualizes the results of the belief that the closer to the purchase the customer was, the more valuable that “closer” channel was), and “Position Based” (where we put the most value on the first interaction and last interaction, and the rest, equally, among all the steps in between).

In the example below, we can see how the “value” of the contribution of each channel changes for this business, when we compare the traditional “Last Interaction” model against the “Position Based” one.

Google Analytics Attribution Model Comparison Tool

Google Analytics Attribution Model Comparison Tool


Keep the Conversation Going

This process of discussing the customer journey, revealing the touch points, and evaluating attribution models will, or at least should, never end. It can be a daunting thing to start, but starting the discussion is the most important step to evolving the understanding of attribution, and constantly improving.

Point It About the author
No Comments

Leave a Comment: