The best model for a business to use is based on factors such as the length of the sales cycle and the size of the purchase.
What is attribution modeling?
An attribution model is a set of rules that allocates credit for sales or leads to touch points in conversion paths.
In later sections, we’ll discuss the specific types of attribution models and which one is right for you. For now, understand that in today’s online world, it’s not common for users to convert on the spot. Instead, it may involve multiple visits to the website, interactions with your different social platforms or newsletters and a number of display ad impressions on 3rd party websites before a purchase is made.
Attribution models are important to understand because depending which model you choose, you will be allocating conversion credit differently to your various marketing channels.
Attribution Modeling and the User Journey
For example, I was recently looking to purchase a new stove. Being brand-agnostic, my search started in Google with the phrase “gas range.” After scrolling through a few Google Shopping Ads, I found one I liked and clicked into the website, read some reviews and left.
A couple of days passed and I wanted to show my wife so I searched for “Frigidaire gas range” and selected the organic listing to the website.
Later that week, as I’m scrolling through Facebook, I see a dynamic product ad in Facebook highlighting a promotion so I click into the website to learn more.
Feeling I made my decision, I search for “Frigidaire [model number]” selected a paid search ad and eventually purchased my product on the website.
To recap, this was my user journey.
As Fridgeaire’s marketing department, do you attribute my purchase to your Google Shopping efforts since that is what first made me aware of their ranges? Does paid search deserve 100% of the credit because it was the converting visit? Do they all deserve a % of credit?
Let’s explore different attribution models before we can answer the questions above.
What types of attribution models exist?
Within Google Analytics, there are predefined attribution models you can use to evaluate your marketing efforts behind different lenses. Those models include:
- First Interaction
- Last Non-Direct Click
- Last Interaction
- Time Decay
- Position Based
The First Interaction model assigns all credit for the sale to the first touch point the customer used, which would be the Shopping Ad in the example above.
The Last Non-Direct Click model assigns all credit for the sale to the last channel that the customer clicked on before converting and ignores direct traffic. In this example, 100% of the credit is given to Paid Search. If my last converting visit involved me typing the website URL directly into my browser (aka “Direct”), Paid Search would still get the credit. Direct would only get the credit if that was the only way I accessed the website. Nearly all GA reports use this model as the default and can’t be changed.
The Last Interaction model assigns all credit for the sale to the last touch point, regardless of the inclusion of direct visits like we saw above. Again, this would grant all credit to Paid Search. GA uses this model to view multi-channel funnel reports by default.
The Linear model assigns equal credit to each touch point in the conversion path. In this example, the Shopping ad, Organic Search, Facebook Ad and Paid Search would all get 25% of the purchase price.
The Time Decay model assigns credit according to how closely in time the touch point is to the sale. This model would assign the most credit to the Paid Search channel, followed closely by the Facebook Ad. The Organic Search channel would receive less credit, and the Shopping Ad would get significantly less credit since it was the furthest channel from the purchase.
The Position Based model gives 40 percent credit to each of the first and last interactions, and distributes the remaining 20 percent credit evenly among the other interactions. In this example, the Shopping Ad and Paid Search channels would each get 40 percent credit, while the Facebook Ad and Organic Search would each get 10 percent credit.
GA also includes a custom modeling tool that allows you to create your own attribution model based on the First, Last, Linear, Time Decay and Position Based models. You can then add in the additional factors that are important to your business. This process requires you to develop a solid understanding of the base models, especially Time Decay.
You then need to determine the user behavior you value most, often by reviewing the historical repeat purchase behavior. There may also be a conversion window that’s ideal for your business, including micro-conversions defined by engagement goals. These answers will provide the context you need to customize your own attribution model.
What model is right for me?
The best GA attribution model to use largely depends on your business model and advertising objectives. For example, the First Interaction model is best if your most important goal is building brand awareness. That’s because the interactions that initiate the conversion are the best indicators of how well a channel is building your brand. Players who are new to their niche often use the First Interaction model, since they typically have less brand awareness than their competitors.
Benefits of Each Attribution Model
The Last Non-Direct model measures traffic sources and marketing channels rather than the actions of potential customers such as directly accessing your site. Use this model instead of the Last Click model if you want to credit the last interaction without giving credit to direct traffic.
Use the Last Touch model when your buyers give minimal consideration to their buying decision, making the last interaction before the sale the most important. This model is common for fast-moving consumer goods where buyers don’t care much about the brand.
The Linear Attribution model is most beneficial for business models in which different interactions with customers are equally important. This model is common for customer service and support, where the customer routinely interacts with staff members multiple times for a particular issue.
Use the Time Decay model to understand your customers’ buying behavior during time-sensitive promotions, as the interactions that occur closest in time to the sale are the most important for this type of marketing strategy. This model is also a good choice when data-driven modeling is unavailable.
The Position Based model is common for advertising objectives in which the middle interactions are less important than the first and last interactions before a sale.
Need some assistance setting up your attribution modeling?
Marcel Digital is a Google Analytics Certified agency, staffing an entire team of Google Analytics Certified experts. We've handled implementations of all sizes from small to enterprise. If your team needs help evaluating your business or needs a custom attribution model in Google Analytics, give us a shout - we're always happy to help!