An Introduction to Google Analytics Attribution Models

Digital Marketing is a combination of data, analytics, and intuition. Computers can’t read minds, however, they can track user touch points. Digital marketing experts use these touch points to better understand the buyer’s journey and learn what drives the decision to buy. If Google Analytics attribution models are new to you, don’t fret! This is only the beginning of our attribution series.

What is attribution?

Picture this — you find an ad for a great pair of shoes on Facebook. You click on it and browse the website. After viewing the product for a few minutes, you exit the site without making a purchase.

Over the next few months, the ad continues to pop up on your feeds because you keep engaging with the content — on Facebook, Instagram, YouTube, and so on. You even sign up for their newsletter. Eventually, in about 3 months, you look up the brand on Google, click on the ad, and make a purchase.

Your conversion path looks something like this: Social Network > Display > Direct > Email > Social Network > Organic Search > Paid Search. All these channels are your touch points.

With tools like Google Analytics attribution models, we can assign credit to various touch points along a conversion path. This is how it looks in Google Analytics:

Why attribution modeling is important

Attribution models help businesses understand how their online marketing channels contribute to their return on investment. They also help marketing teams analyze which channels have a higher cost per acquisition (CPA), which channels to optimize, and which channels have the best return on investment.

This can help companies use their advertising dollars effectively, and focus on channels that drive the most conversions.

Most small businesses don’t use Google Analytics attribution models to their advantage. Using the right attribution model is empowering. It can help set you apart from your competition because it gives you deep insights into user behavior.

Tools Google Analytics offers

Google Analytics offers several attribution models and a pretty sweet Model Comparison Tool which allows you to compare data between two different attribution models.

It’s important to mention here that these models are not perfect and may not have all the answers that you are looking for. They each have pros and cons, which we will discuss in subsequent posts, as we dive deeper into each model.

Last Interaction

The Last Interaction model is self-explanatory. It gives 100% of the conversion credit to the last touchpoint. In our example above, the last touchpoint was Paid Search. Last Interaction will assign 100% credit to Paid Search, which is flawed because it ignores every other channel that contributed to the conversion at different points.

Someone looking at this data would then see a higher percentage of conversions coming from paid search, and a low percentage coming from social ads. This can mislead someone to believe social ads are not performing well for them, when in fact they are.

We will dive deeper into Last Interaction in our next post. Stay tuned!

Last Non-Direct Click

The Last Non-Direct Click model ignores all direct channels. When users land directly on your website by typing your website’s URL into a search engine, the Last Non-Direct Click model assigns 100% of the credit to the last channel the customer converted from. In this case, it would again be Paid Search.

There are situations where this could work. If your online store is well known, you could potentially use this model to determine how all paid channels are contributing towards that final conversion action.

Last Google Ads Click

100% credit is assigned to the Last Google Ads Click. Taking our example for the shoes again, 100% credit goes to the Paid Search channel, ignoring every other conversion path.

When using the Model Comparison Tool, this could be a helpful exercise so you have more insight into how Google Ads are leading to conversions. But overall, this is definitely not a great model to rely on.

First Interaction

The First Interaction model assigns 100% credit to the first click. In our shoe example, this would be Social Network. While it is true that the user journey originated through social ads, it was not the only touchpoint.

Linear

The Linear model assigns equal credit to each touchpoint. Let’s consider a conversion path that looks like this:

Direct > Referral > Email > Organic Search

Each channel here gets 25% credit for the conversion.

This model is flawed as well because it is unable to assign an actual weight to how each channel contributed towards the conversion. If the Direct touchpoint was in April and the Organic search touchpoint was in September, did Direct really contribute 25% to the conversion? Definitely not.

Time Decay

A personal favorite, Time Decay, despite its shortcomings, is one of the best models available in the free Google Analytics version because it assigns a higher credit to touch points that occur closer to the conversion.

Let’s go back to the conversion path for shoes: Social Network > Display > Direct > Email > Social Network > Organic Search > Paid Search.

Google Analytics will assign the most credit to Organic and Paid Search while assigning lesser credit to Display and Direct.

Position Based

With the Position Based model, 40% credit is assigned to first and last interaction each, and 20% credit is distributed to all other channels. In our example, 40% credit would go to Social Network, 40% to Paid Search, and 20% would be distributed to everything else in between.

This is a rather clever model since it assigns a higher percentage of credit to the first and last interaction. The conversion path would never have begun if the first interaction did not happen. Additionally, it would not have resulted in a conversion without the last interaction.

Conclusion

Google Analytics attribution models offer a unique insight into the mind of your customer. Conversion paths are not the final frontier when it comes to understanding user behavior since each user can follow a completely different path depending on the channels they interact with.

Over the next few months, we will dig deeper into each individual attribution model and analyze their advantages and disadvantages.