About attribution modeling

An attribution model is the rule, or set of rules, that determines how credit for sales and conversions is assigned to touch points (impressions and clicks) in conversion paths. For example, the Last Interaction model assigns 100% credit to the final touch points that immediately precede sales or conversions. In contrast, the First Interaction model assigns 100% credit to touch points that initiate conversion paths.

In Campaign Manager 360, you use the Attribution Modeling Tool (AMT) to compare the different models of assigning credit to channels that are discussed in this article. This article provides the background information that you need to understand attribution modeling. To learn how to run and compare these models in Campaign Manager 360, see The Attribution Modeling Tool (AMT).

Default attribution models

The Attribution Modeling Tool provides the following default attribution models. You can also create your own custom models in the tool.

The Floodlight model attributes 100% of the conversion value to the last click made by the user before buying or converting. If there was no click, the model attributes the value to the last impression. Learn more about Floodlight conversion counting

Use case: This model matches the conversion attribution shown in standard reports in Campaign Manager 360. It's useful for aligning attribution with Floodlight reporting.

 

The Last Interaction model attributes 100% of the conversion to the last channel
with which the customer interacted before buying or converting.

Use case: This model emphasizes the role of a channel in closing sales or conversions. It's relevant for campaigns that are designed to attract people at the moment of purchase or conversion. It's also useful if your business is mainly transactional, with a sales cycle that doesn't involve a long consideration phase.
 

The First Interaction model attributes 100% of the conversion value to the first channel with which the customer interacted.

Use case: If you're running campaigns to create awareness. For example, if you're marketing a brand that's not well known, you might put a premium on keywords or channels that first expose customers to the brand. Use this model to emphasize the role of display advertising in initiating conversion funnels.
 

The Linear model gives equal credit to each channel interaction on the way to a conversion.

Use case: For campaigns that are designed to maintain customer contact and awareness throughout the sales cycle. In this model, each touch point is equally important during the consideration process.
 

The Time Decay model is appropriate for a sales cycle that involves only a brief consideration phase. This model is based on the concept of exponential decay. It gives the most credit to the touch points that are nearest to the time of conversion or sale. The Time Decay model has a half-life of 7 days, meaning that a touch point 7 days before a conversion will get half the credit of a touch point on the same day as the conversion or sale. Similarly, a touch point 14 days before the conversion will get 1/4 the credit of a day-of-conversion touch point. The exponential decay continues to the end of your lookback window.

Use case: For short-lived promotional campaigns. If you run one-day or two-day promotions, you might want to give more credit to interactions during those promotions. Touch points further back in time have less value compared to those that occured right before the conversion.
 

The Position Based model is a hybrid of the Last Interaction and First Interaction models. Instead of giving all the credit to either the first or the last interaction, you can split the credit between them. A common scenario is to assign 40% credit each to the first and last interaction, and assign the remaining 20% to the remaining interactions.

Use case: When you value the touch points that introduce customers to your brand or promotion, as well as the touch points that result in sales or conversions.

 

 The Social model is based on the linear model but impressions are weighted to account for social interactions. The default weightings are:

• Impressions without social engagements = x0.5
• Impressions with any low-value social engagements (and no high-value) = x0.75
• Impressions with any high-value social engagements = x1.5

High-value social engagements extend reach to other users. Examples include shares, and retweets. Low value social engagements show interaction, but don't extend reach to other users. Examples include expands, and profile views.

 

Use case: When measuring campaigns run on social networks like Twitter.

Attribution modeling example

A customer finds your site by clicking one of your Rich Media ads. She returns one week later by clicking on a search ad. That same day, she comes back a third time via one of your email campaigns (which uses a click tracker), and a few hours later, she sees a standard ad while booking some holidays and returns again directly to make a purchase.

In the Floodlight attribution model, the last click—in this case, the Click Tracker channel—would receive 100% of the credit for the sale.
 

In the Last Interaction attribution model, the last touch point—in this case, the Standard Display channel, via the impression of the standard ad on the booking site—would receive 100% of the credit for the sale.
 

In the First Interaction attribution model, the first touchpoint—in this case, the Rich Media channel—would receive 100% of the credit for the sale.
 

In the Linear attribution model, each touchpoint in the conversion path—in this case the Rich Media, Search, Click Tracker, and Standard Display channels—would share equal credit (25% each) for the sale.
 

In the Time Decay attribution model, the touchpoints closest in time to the sale or conversion get most of the credit. In this particular sale, the Click Tracker and Standard Display channels would receive the most credit because the customer interacted with them within a few hours of conversion. The Search channel would receive less credit than either the Click Tracker or Standard Display channel. Since the Rich Media interaction occurred one week earlier, this channel would receive significantly less credit.
 

In the Position Based attribution model, 40% credit is assigned each to the first and last interaction, and the remaining 20% credit is distributed evenly to the middle interactions. In this example, the Rich Media and Standard Display channels would each receive 40% credit, while the Search and Click Tracker channels would each receive 10% credit.