Google's Journey into Machine Learning: What Marketers Need to Know
With the new Google Analytics 360 Suite, machine learning comes to center stage as marketers contend with how to use this technology in their digital strategies. The recently launched enterprise marketing cloud suite is working to bring together the once siloed data of disparate Google products. Just before the launch, eMarketer’s Jillian Ryan sat down with Google analytics evangelist Justin Cutroni to discuss how Google approaches machine learning and what these advancements mean for marketers using Google products.
eMarketer: What is Google’s approach to machine learning and how it is used in the Analytics platform to benefit marketers?
Justin Cutroni: At Google we’re enamored with machine learning and investing heavily in it, and that manifests itself in a lot of our products. We’re really interested in leveraging machine learning, so that businesses can take more action on the data. There are marketing plans and economic forces, so teaching a machine to really understand all of those nuances is challenging, but what we have been trying to do is look at how we implement machine learning at a very basic level.
“Machine learning helps identify what conversions are more high-value.”
eMarketer: Predictive models and predictive marketing are generating a lot of buzz at the moment. Is Google implementing machine learning to employ predictive strategies for companies using Analytics?
Cutroni: Machine learning helps identify what conversions are more high-value. Within Analytics, Smart Goals is a feature that brings in different metrics like session length, page views, geographic location, device and so on. We can look for signals that impact those metrics and use that information to say a certain subset of users are more valuable because they’re different from all of the other visits.
The sessions and the visits to a business’ website are not all of equal value, and a company shouldn’t be paying the same amount for all leads. Rather they should be marketing to the people who are going to have a high propensity or likelihood of doing what you want them to do. Then we let the machine figure that out so marketers can optimize around it. This has always been hard, but machine learning is helping us get across that barrier.
“The sessions and the visits to a business’ website are not all of equal value and a company shouldn’t be paying the same amount for all leads.”
eMarketer: These are all top-funnel approaches, getting new and qualified leads into the system. But what about more mid-funnel implications and predicting things that churn, and retention rates?
Cutroni: Now you’re getting into what I’m really excited about: breaking down organizational silos and helping businesses integrate their data sources, an incredibly important piece of which is CRM data. We have the machinery to connect the data to drive new customer acquisition and also reduce churn, helping our users better leverage their existing customers and retain them for longer periods of time.
eMarketer: What do you see on the horizon for the future of machine learning and its application for marketers?
Cutroni: It’s big data, it’s customer analytics, it’s visual analytics; with machine learning the data works together. We will go beyond reaching a certain type of person with a specific message on this type of device. Because once a user performs a trigger behavior, we understand that they are more qualified and marketers can then change that message and adjust offers, frequency and advertising more efficiently.
We can really focus on using data and information to increase the relationships that consumers have with businesses. We’re definitely not there yet—we don’t have all of the things connected the way we want them to and we don’t have the machine running, working across all of the data, but we have all of those pieces and it’s going to be an exciting ride to get there.