Manage attribution models

About attribution models

An attribution model determines how to distribute credit for a conversion across all of the activity in a conversion funnel. By default, DoubleClick Search (DS) uses the last-click attribution model, which gives all credit to the last click that occurred before the conversion.

If you want a Floodlight activity column, Google Analytics activity column, or DS bid strategy to give some amount of credit to clicks on paid search ads higher up in the funnel, you can use an alternate attribution model.

Attribution models are defined in DoubleClick Campaign Manager (DCM), Google Analytics, or DoubleClick Search. To use a model, a DS agency manager or advertiser manager adds or creates it in DS. DS can then sync historical performance data with the model as well as apply the model going forward.

The agency manager or advertiser manager can add up to 5 attribution models from DCM, up to 5 attribution models from Google Analytics, and create up to 5 data-driven attribution models in DoubleClick Search for a total of 15 models.

Note that:

  • DS uses the last-click attribution model by default. The DoubleClick time decay model, an alternate attribution model, has been added for you so that you can start using it immediately.
  • Even if you're using other alternate attribution models in DCM or Google Analytics, you can't use the models in DS until an agency manager or advertiser manager adds them to DS.  
  • Only one type of attribution model, the data-driven attribution model, is created in DoubleClick Search. The other alternate attribution models are added to DoubleClick Search from DCM or Google Analytics.
  • Floodlight activity columns and bid strategies can use attribution models added from DCM or created in DoubleClick Search.
  • Google Analytics activity columns can use attribution models added from Google Analytics.

Pre-defined models

Both DCM and Google Analytics provide a set of basic attribution models that you can add to DS. In addition, you can create custom models in DCM or Google Analytics and add them to DS.

Basic attribution models

Both DCM and Google Analytics provide a common set of basic attribution models that you can add to DS: 

  • Time decay: Most heavily credits the interactions that occurred nearest to the time of conversion. The half-life for the decay is 7 days. That is, clicks that occur 7 days before the conversion receive half the credit, while earlier clicks receive less than half, and later clicks receive more than half. Use this model if your ads and campaigns are typically short-lived and promotional. The DoubleClick time decay model has been added for you.

  • Linear: Gives equal credit for the conversion to each interaction click in the conversion path. For example, if a consumer clicks on a display ad, a paid search ad, and a natural search result before a conversion, a Floodlight or Google Analytics activity column that shows actions will credit the keyword responsible for the paid search click for 33% of the action. If the column attributed 200 actions to a keyword before the conversion, it will show 200.33 actions after the conversion. Use this model if your goal is for your ads and campaigns to maintain contact and awareness throughout the sales cycle.

  • First interaction: If the first click in the funnel to a conversion is on a paid search ad, then that click receives 100% of the credit for the conversion. If the first click is from a display ad or a natural search result, the conversion is not reported.

  • First paid search interaction: Attributes 100% of the credit for a conversion to the first click on a paid search ad. Even if a customer clicks on a display ad or a natural search result before clicking a paid search ad, all credit is given to the paid search ad. If there are no paid search clicks in the funnel, the conversion is not reported. That is, if a funnel only contains clicks on display ads and natural search results, DS does not report a conversion. Use this model to emphasize the role of paid search clicks in initiating conversion funnels.

  • Last paid search interaction: Attributes 100% of the credit for a conversion to the last click on a paid search ad. Even if a customer clicks on a display ad or a natural search result after clicking a paid search ad, all credit is given to the paid search ad. If there are no paid search clicks in the funnel, the conversion is not reported. That is, if a funnel only contains clicks on display ads and natural search results, DS does not report a conversion. Use this model to emphasize the role of paid search in closing conversions, or as a basis of comparison with other models.

  • Position based: Most heavily credits the first and last interactions in the conversion path and gives less credit to any interactions in between. Specifically, this model attributes 40% of a conversion to the first click, 40% to the last click, and distributes the remaining 20% evenly among the clicks in between. Use this model if you value the role of paid search in initiating conversions but also want to assign significant credit to the channels that resulted in a conversion.

  • Assists: Attributes the credit evenly among all clicks except the last click. Use this model if you place the highest value on all the clicks that assist in a conversion.

Note that DCM models can only be used with Floodlight columns, and Google Analytics models can only be used with Google Analytics columns. For example, if you wanted to use a Linear model in both Floodlight activity columns and Google Analytics activity columns you'd need to add both the DCM linear model and the Google Analytics linear model.

The Google Analytics Last AdWords Click and Last Non-Direct Click attribution models cannot be added to DS.

Scaled conversion data for Floodlight attribution models

The basic attribution models from DCM use scaling to account for the conversion data that can't be directly measured. Learn more

Custom attribution models

You can create custom attribution models in DCM and Google Analytics and then add them to DS. Some of the options and rules you can specify in DCM and Google Analytics are not compatible with DS.

Custom attribution models do not include scaled conversions. Learn more

Before you create custom models, see:

DS also supports Adometry and other attribution models

DoubleClick Search supports Adometry and other third-party attribution models. Contact your DS representative for more information.

Syncing 60 days of historical performance

When you add an attribution model to DS, DS applies the model to the previous 60 days of performance data, as well as to all performance data going forward. Because DS needs to wait until performance data for the current day is complete and available, the sync may not finish until the next day.

Data-driven attribution models

DoubleClick Search can analyze real data that's unique to your search campaigns and use the data to generate a data-driven attribution (DDA) model. The DDA model learns how to distribute credit for conversions across all of the interactions in a conversion path based on actual interactions that are driven by your search campaigns. (DCM can also generate a data-driven attribution model, but the model you create in DCM is completely separate from DS models. It cannot be imported into DS, so it can only be applied to performance data in DCM.)

Data driven attribution use scaling to account for the conversion data that can't be directly measured. Learn more

Learn more about DDA models in DS.

Attribution models and impressions

Attribution models always ignore impressions of search ads. This applies to the basic models and to custom models in DS and in DCM.

In DS, the basic models and any custom models that you create for use in DS ignore impressions of display ads. However in DCM, the same basic models do include impressions of display ads, and you can create custom models that include display impressions. This means that reports in DS and DCM may not match. For example, a DS report that uses the Linear attribution model does not attribute credit to display impressions. A report in DCM that uses the Linear attribution model does attribute credit to display impressions.

Was this article helpful?
How can we improve it?