Attribution Models 101: Use Data to Maximize Your Marketing Budget

Even if you have a general idea of how effective each marketing channel is to your business’ success, wouldn’t it be better if you could look at your data, analyze the results, and know which platform or touch point is giving you the highest ROI? Or the lowest?

Imagine how access to that information would impact your marketing budget, your team’s workload, your ROI, and your sanity!

That’s exactly what attribution modeling can do for your business.

Google defines attribution as “…the rule, or set of rules, that determines how credit for sales and conversions is assigned to touchpoints in conversion paths.”

In this article, we’ll be walking you through the most common attribution models and review the impact they could have on your marketing efforts.

For the sake of clarity, we’ll be using the following example to explain each attribution model and will refer to it throughout the article.

Example Customer Scenario

A customer discovers your website by clicking on an ad they saw while browsing the web. While they are there, they opt in to receive emails from your brand. Later on that week, they open an email from you and return to your product page. The next day, they find you on social media and like your page. Two hours later, they visit your site directly and buy your product.

The Last Interaction Attribution Model

This model gives 100% of the credit to the very last channel that the customer interacted with before buying. From the above example, that would mean that direct traffic to your website would get 100% of the credit for that customer converting. (As a reminder, a direct channel is when the user types the URL into the browser directly.)

This model is standard in almost all analytics tools, except for Google Analytics. This model works when the sale cycle is very short. However, for most ecommerce websites, this model will not give you the insights or clarity you need to optimize your ad spend or your marketing efforts in the long term. Therefore, we do not recommend this model at all.


The Last Non-Direct Click (“Last Touch”) Attribution Model

In this model, 100% of the credit is given to the last non-direct channel that the user interacted with. If the last channel is direct, the conversion is attributed to the last campaign or marketing activity the user had, whether it is organic search, email, social, referral, etc.

In the example scenario, the conversion would have been assigned to the email campaign.

This model is the default in Google Analytics reports and it can not be changed (except for the Model Comparison Tool, which we’ll talk about later).

The downside of this being the default model in Analytics is that it underestimates value of direct traffic; any brand recognition metrics are instead attributed to the last marketing campaign activity.


The First Interaction (“First Touch”) Attribution Model

As we’re sure you can infer from the name, in this attribution model 100% of the credit for a sale would be attributed to the very first channel the customer interacted with. From the example, the web ad the customer clicked would get credit for this sale.

This may help you detect which channel is bringing new potential customers as well as its conversion rate, but the first touch attribution model doesn’t take into account the various steps the customer took and doesn’t provide full context for the resulting sale.


The Multi-Touch Attribution Model

While the first and last touch attribution models are easy to understand and used by most analytics software, they’re not nearly as useful as multi-touch attribution.

Multi-touch attribution takes into account all the various touch points that the customer interacted with leading up to a sale, and provides each touch point with a value or weight.

So, how do you know what values to assign each touch point? Good question! There are different ways to allocate the value of each touch point, explained below:


This model takes into account each of the touch points the customer has interacted with and says they all had the same impact on the resulting sale. But if your intentions for using attribution are to streamline your marketing efforts or to understand where your best results are coming from, using this method obviously won’t solve the problem.


Time Decay

This method assigns more weight to the touch points that were closest to the sale. While this model acknowledges and gives credit to the first click that caused the customer to go to your website, it will give each interaction more credit for the sale the closer you get to the sale taking place.

Analytics expert and former Director of Research & Analytics at Intuit, Avinash Kaushik advocates the use of the time decay model saying:

“You only have to think about it for five seconds to realize it passes the ultimate test for everything.”


Other attribution models are:

  • The Last AdWords Click: This gives 100% of the credit to the last AdWords Ad that the user interacted with. Not very useful when you are studying the impact of all your marketing efforts.
  • The Position-Based Model: In this model, 40% of the credit is assigned to each first and last interaction and the remaining 20% is divided for the touch points in the middle.

Model Comparison tool

As we mentioned, Google Analytics reports use the Last Non-Direct Click, except for one amazing tool that allows you to compare your conversions with different models.

You will find this on your Analytics Platform in Conversions > Attribution > Model Comparison Tool.

In this report you’ll be able to compare up to three different models and see how each channel performed. Below is an example of a report comparing First Interaction, Last Non-Direct Click, and Time Decay.

comparison model

On the left are all the channels that had a conversion in the period of time selected, and in the center we see, sorted by each model attribution, the Conversions and the Conversions Value.

From this example we see that in the First Interaction Model the Direct Channel has the majority of conversions, which tell us it is a well-recognized brand.

On the Last-Non Direct Click, we also see that Direct has the most conversions. If we compare it to the First Interaction Model, we see that users are converting on their first and only visit. Otherwise, the conversion would be assigned to another channel in the second model.

By this assumption, it’s logical that in the Time Decay model, Direct Traffic has most of the conversions since it’s a short customer journey and may be the only channel they interacted with.

The other channels have similar behavior, except for Referral, which doesn’t seem to be converting in the First Interaction. With this data, it could be a good idea to review the sources that are driving Referrals and check if they are linking to a page that motivates conversions in line with the other channels.


This is an example of some of the insights you can take away with this report. It’s always good practice to investigate and compare different models and find out which one suits your business best.

Attribution modeling for your digital advertising spend is not an easy task, but it’s very worthwhile. Any ecommerce manager or online marketer looking for an edge needs to be actively implementing attribution models into their marketing efforts.

While your competitors are throwing spaghetti against the wall and hoping that something will stick, you can have data-driven marketing projections and a clearer understanding of how your marketing dollars should be spent.

Related Posts