October 14, 2021
Data-Driven Attribution is Coming to Google Ads
Google Ads metrics are just random numbers if the data isn’t actionable.
Google knows this, which is why they are always tinkering with the way data is measured and collected.
For some time, Google Ads has been using a last-click attribution model (a rules-based model) as the default to determine conversion for many accounts. Last-click by default is soon coming to an end.
Google recently announced it will be moving all Google Ad accounts to default to their data-driven attribution model.
Here’s just a little bit of information on what that means and why it’s so important to really make the most out of Google Ads data.
What is data-driven attribution?
We first mentioned last-click attribution, so let’s explain how that differs from data-driven attribution. Last-click attribution records the immediately preceding ad a customer clicks on before converting. While this is nice to know, it doesn’t really tell you what prompted a visitor to turn into a prospect.
Data-driven attribution, however, looks at more than just that ad clicked. It examines your website, your online store (if applicable), your ads, and any other factor someone can experience that might make them decide to purchase. Data-driven attribution also examines the buyer’s journey through your marketing. It also assigns more credit to those ads that really had the most impact, even if they weren’t the last ad clicked.
As its name suggests, data-driven attribution needs data to work properly. That’s why in the past, this has only been available to those accounts with a high volume of ad interactions and conversions. All that is changing now that Google has been supercharging data-driven attribution with machine learning.
Basically, machine learning can extrapolate and interpret interactions with even a limited dataset. It lets even small companies use this great tool for examining the true story behind their metrics. As the new default setting, data-driven attribution will provide that machine learning with more data to digest. And the more data the machine learning can ingest from all of us, the more accurate its inferences will be.
Plus, at least for now, though data-driven attribution will become the new default, five rules-based models will also be available. This includes last-click, since so many small and medium-sized businesses have become used to that model.
Why is data-driven attribution so important?
Apart from soon being more available to more businesses, data-driven attribution is important because it gives us more insight into which ads are performing the best. The easiest way to explain this is through an example (building off the example given by Google in an older page explaining this attribution model).
Data-driven Attribution Example
You are running ads for a new home community (let’s call it Homeland) attracting traffic to your website to get potential buyers to call and schedule a tour. Currently, you are running three different ads:
Ad 1. “Home is where your heart is.”
Ad 2. “Come home to Homeland.”
Ad 3. “You can’t spell “Homeland” without “home!””
It’s well known that multiple touchpoints happen before conversion (e.g., multiple ads are clicked before someone decides to take that next step). This example is no exception. Data-driven attribution identifies that customers who click on the Ad 3 followed by Ad 2 are more likely to convert than those who click on either ad alone.
It also looks through every other combination of ad chains leading to conversion and realizes they all have one thing in common: Ad3. It thus determines that Ad 3 gets more credit for each conversion, which is reflected in your analytics.
So what did we learn from that example? One of your ads might be the last one clicked before conversion, but it may not be the most important ad. This is in regards to convincing a customer to take the final step and vice versa. This knowledge gives you a much clearer window into how your ads are really performing.
We also have to mention that because machine learning has evolved to make logical inferences based on small quantities of data, it’s a great solution for adapting to the cookie-less future that is on the horizon. Data-driven attribution’s machine learning can use its aggregated data to fill in the holes left by the lack of cookies. Eventually, this might even prove more effective than cookies.
What to expect
Google plans to roll out data-driven attribution as the default beginning in October to select accounts. The full switch will be over by early 2022. It will be interesting to see if data-driven attribution lives up to Google’s claim of helping “every advertiser clearly understand the full value of their Google Ads campaigns.”
At Ai, we’ve always believed that there are no “silver bullets” or single steps that cause a prospect to become a customer. Not one ad or even one platform. It is the collective work of all our marketing efforts that educate and persuade prospects to convert. The clarity of data-driven attribution should hopefully help us better quantify our marketing philosophy and show that a rising tide of many touchpoints lifts all our efforts.