Get More Insights from Google Ads Data Modeling
By Graham Hooten

January 11, 2023

By Graham Hooten

January 11, 2023

Google has recently started automatically switching certain Google Ads campaigns from a last-click attribution model to data-driven attribution. Advertisers have the option of opting out of this switch, but with data-driven attribution becoming the default, questions have to be answered on what the impact will likely be and whether to move forward with the automatic switch on a given account.

What Is Data-Driven Attribution?

Data-driven attribution is a model that gives credit throughout the conversion journey, which allows you to see all of the ad interactions that contributed to a conversion rather than last-click attribution that gives full credit only to the last data source. This means that multiple keywords, ads, ad groups, or campaigns could receive partial credit in the form of decimal amounts rather than a whole numeric value for a given conversion. The methods of interactions that would give credit for a conversion include clicks or video engagements within Google Ads.

Attribution Variance Depends On Campaign Structure

As with any decision about campaign goals, strategies, or reporting methods, data is going to be one of your finest teachers. We’ve looked at the results from several campaigns that have either made the switch to data-driven attribution, or utilized it from the beginning.
In one account in the food & beverage sales industry, we saw 20% of Search keywords receive partial credit for a conversion after data-driven attribution was implemented. We also saw a 16% increase in new keywords receiving partial credit for a conversion that did not receive any credit the previous period. This account has Search campaigns split by region, so the impact in attribution change was seen at the keyword level.
Another account in the environmental products and software industry has been using data-driven attribution since campaign launch, and has multiple non-brand campaigns and ad groups organized by product. Each campaign has had more than 50% of the keywords that received credit for a conversion be given partial credit, which indicates even within Search only, data-driven attribution can illustrate multiple touchpoints in the path to conversion if you have multiple related products. These insights are immensely helpful as you review keyword performance, search term reports, and your campaign structure.

What Conclusions Can Be Drawn?

Based on the above examples, impacts from data-driven attribution can give you greater insight into keyword performance, and learnings can be expanded to the campaign level when you run several campaigns that lend to multiple touch points and have a variety of products or services that the same customer could be interested in. If your account is in the financial services, home goods, and other fields that fit this description, you could be in business choosing this attribution method.
Another factor to consider is your channel mix. In addition to Search, are you also running Display, Performance Max or YouTube campaigns? Data-driven attribution shows added value for top of the funnel and middle of the funnel tactics more than than last-click attribution, so testing this model could make a lot of sense on those accounts. Instances where DDA may not be the ideal model, or you won’t see much benefit from, are campaigns with a small volume of conversions each month. Any attribution model is going to give greater insight with more data points, but this is especially true of DDA.

If you’re looking for a team armed and ready to manage your campaigns with an intelligent, thoughtful approach to data and performance, let us know! We’re happy to talk through how we can elevate your business together.

Written By Graham Hooten

Graham is on the ad operations team at Arm Candy with a particular passion for programmatic and social channels.

Written By Graham Hooten

Graham is on the ad operations team at Arm Candy with a particular passion for programmatic and social channels.