Data Driven Attribution Unraveled

Attribution has been one of the central problems for marketers since the beginning of analytics. How much credit should we attribute to a specific channel? Although there is no one-size-fits-all answer, everyone agrees that the default Last Non-Direct Click Attribution in Universal Analytics is far from ideal.

In January 2022, Google attempted to aid this problem and introduced Data-Driven Attribution (DDA). However, the reports for Data-Driven Attribution are available in GA4 for the dates starting November 1st, 2021, assuming you set up your GA4 property before this date.

So what is Data-Driven Attribution, and how is it different from the Last Non-Direct Click Attribution of Universal Analytics?

Google’s definition of Data-Driven Attribution:

“Data-Driven Attribution distributes credit for the conversion based on data for each conversion event. It’s different from the other models because it uses your account’s data to calculate the actual contribution of each click interaction.

Each Data-Driven model is specific to each advertiser and each conversion event.”

The upside of the Data-Driven Attribution is that it attempts to assign a more accurate credit to each channel on the conversion path, excluding the Direct channel unless it’s the only channel on the conversion path. Instead of using a set of rules, e.g., using only the last or first channel on the conversion path, it distributes the credit for conversion among the channels based on how likely they were to contribute to this conversion.

In practice, you may see only a slight difference in conversions between DDA and Last-Click Attributions when you first start with GA4.

Similar to the Last Non-Direct Attribution, DDA attributes credit to Direct only when it’s the only channel on the conversion path. That’s why you should see the same number of conversions attributed to Direct for both attribution models in the Model Comparison Report.

When we compared DDA to the Last-Click model, the results varied from client to client, but overall, the DDA seemed to favor Paid Search more than the Last-Click model, while Organic Search had a slightly lower number of conversions.

Things to keep in mind:

  1. The attribution model in GA4 is set to DDA by default, so you don’t need to change any settings if you prefer to use this model.
  2. If you prefer not to use DDA in your GA4, you can change your default attribution model to any other model, which will be reflected in all GA4 reports affected by the attribution model. 
  3. If you change the default model in your GA4 property, these changes will be retroactive, so you can view your historical data using this model.
  4. Be mindful of the dates when the attribution models were introduced. The DDA data is only available in GA4 starting November 2021, while the rules-based attribution models have been available since June 2021.

The main drawback of the DDA model is that it’s a black box, so we don’t know the exact method used to distribute the conversion credit. However, using the attribution model based on the probability of conversion sounds far more compelling than using arbitrary rules like giving full credit to the last or the first channel on the conversion path.

The GA4 reports are only as effective as the quality of your data. Contact us to set up your GA4 property or to audit your existing tracking.


Nick McNeill

Interactive Director

Nick uses his talents in computer science and graphic design to grow the online presence of brands such as Santee Cooper, Southern Tide, frogg toggs, Farmers Telephone Cooperative and Blue Force Gear. A serial marathon runner, his steely determination shines through every brand he grows. As lead user interface designer of the GuestDesk software suite for online hotel reservations, he watched it explode from $1 million in reservations to over $400 million annually.

Read more posts from Nick