Google’s Data-Driven Attribution: What You Need to Know

By Kimberly Honoré in Media Strategy on May 23rd, 2017

Google Attribution, digital marketing, advertising

First and foremost, if you have a free moment, do yourself a favor and go online and read all about Google’s Data-driven Attribution. Google’s article, “About Data-driven Attribution,” is also a must-read.

As busy marketers, though, we know you don’t always have the time to spend researching the many new and exciting technologies constantly being introduced into the ad landscape. So we’ve summed up Google’s Data-driven Attribution Model in a nutshell. Here’s what you need to know:

  1. Many advertisers are still behind the times

Over 50% of marketers are still giving credit to the last click. But when you move away from this and move towards a data-driven attribution model, you may see up to a 25-50% increase in performance. Also by crediting only the last click, you could possibly be missing out on a whopping 20-40% of your potential ROI, according to IBM’s Global Survey of Marketers.

  1. You need enough data

You need to have at least 15,000 clicks, and your conversion action must have at least 600 conversions tied to it within the last 30 days, for Google to start building a data-driven attribution model.

When (and only when) you have enough data, you will see the data-driven attribution in AdWords. You can go back as far as April 2016 to look at how your historical data looks with data-driven attribution applied.

Also, if you have more than one conversion type, you may see data-driven attribution is available for some of your conversion types, but not all. Again, this is based on whether the necessary benchmarks are there for Google to report on the data.

  1. How it works

Here’s the deal. We’re going to leave out all the fancy advertising and analytics buzz words and try to explain this as easily as possible.

When it comes to converting online, people usually conduct several searches and possibly click on multiple ads on different devices before they actually convert, right? We’ll refer to these as the “paths to conversion.” So Google looks at all the different paths to conversion and finds commonalities between them (using data from converting and non-converting users). From there, Google can determine the actual contribution of touchpoints across the conversion path and assign fractional credit for a conversion based on each touchpoint's propensity to lead to a conversion.

So let’s say that, when looking at paths to conversions for a hotel brand, Google sees that people searching on the term “Cheap Hotels” are 20% more likely to convert then when that query is not part of their path to conversion. Because of this, Google would give more credit (or weight) to that term when it’s a part of a path to conversion—even if it wasn’t tied to the last click.

  1. Start bidding smarter

Once data-driven attribution data is available for your conversion type(s), you can select this model as the source for bidding and reporting in AdWords. From there you can set-up automated bidding in Google’s User Interface (UI) to help adjust bids in accordance with your account goals.

Remember to think of this rollout as a test, however. The end goal is to drive more conversions at a more efficient cost per, or maximize the volume of conversions before you hit your cost per threshold. If that isn’t happening, perhaps a data-driven model isn’t the solution for you. For many advertisers this will be unchartered territory, so make sure to get input from your Google reps and your agency every step of the way.

Here’s more information on what signals Google takes into account for bidding.

  1. There are limitations

For now, this form of attribution is only compatible with Google Search and Shopping campaigns, so data from other engines and channels (i.e. display, email) haven’t yet been incorporated. Also, even when dealing with Search and Shopping campaigns, the model will not work across multiple accounts. So if you have your Search and Shopping campaigns housed in different accounts, Google would only be able to use data-driven attribution for bidding within each account separately. It goes without saying that these limitations alter the overall effectiveness of the model, but our paid search department has been told that Google is moving swiftly to make the model more inclusive in the coming year.

Regardless of any limitations, it makes sense to go ahead and become well-versed on Google’s data-driven attribution capabilities now.  This way you’ll already know how it works and how you can drive improved performance for your campaigns.

Interested in working with our data driven experts?

Connect with our media team now