New streams of customer data should be a blessing to any business. But as VP of marketing for Google Cloud, I sometimes hear from marketers about the challenge of finding actionable insights. As marketing data gets more pervasive and sophisticated, it can be complicated to connect all the data points to basic business goals we all share: acquisition, retention, brand love.
Fortunately, there is new cloud technology that is becoming accessible to marketers, particularly in the form of advanced data analytics and machine learning. Here are three trends we’re seeing from leading marketers who are having early success with analyzing their data in the cloud.
1. Creating a holistic view of customer relationships
Imagine you are leading marketing for a car company and you have millions of customers shopping on hundreds of websites. We studied this last year by tracking a young mom named Stacy, and we found that she made 900-plus digital touch points (71% of them on mobile) before choosing a new car for her growing family. Stacy's 900-plus touch points came from all over—mobile, desktop, and tablet—and they included everything from dealer sites to social media chatter. In a typical organization, the data from these interactions is also widespread between different teams, in siloed channels, using different systems and technology.
But it doesn’t have to be. As a recent study by Google and Econsultancy confirmed, we're seeing that leading marketers are making a serious effort to unify this data and create a holistic view of their customer relationships. Our study showed that leading companies are 1.5X as likely as mainstream companies to have an integrated marketing and advertising technology stack.1
Businesses who want to truly understand consumers like Stacy can use new data platforms, such as the combination of Analytics3 and BigQuery, to bring together the full ecosystem of paid search, display ads, social engagement, dealer site data, inventory, finance, contact forms, and more in one standardized data warehouse.
2. Producing insights with unified data
With connected data, marketers can uncover new insights about customer behavior that weren’t obvious before. A Gallup study recently showed that marketers who use customer data in marketing outperform peers by 85% in sales growth and more than 25% in gross margin.2
Consider how Multichoice, the largest pay-TV company in Africa, used insights to match the customer journey with its own needs.
A content-streaming competitor entered the South African market in 2016, so customer retention was top of mind for Multichoice. Its team needed to understand how to drive long-term engagement and reduce churn with its 8.5 million customers. To create a single view of its customers, the company combined data from 27 different silos—2.5 billion rows and 2 terabytes of data in all—and poured it all into BigQuery, Google Cloud Platform’s marketing data warehouse.
A month later, Multichoice used the data to identify a key insight about its subscriber relationships. It found that customers who purchase many TV channels but watch only a few are more likely to cancel service completely. That led Multichoice to create a new low-cost channel bundle, along with a loyalty rewards program and real-time win-back campaigns. These efforts have helped Multichoice successfully retain existing customers and improve their brand experience.
By unifying data, marketers can uncover new insights about customer behavior that weren’t obvious before.
3. Applying machine learning data to media plans
One of the key advantages of moving data to the cloud is that it enables more complex analytics, increasingly in the form of machine learning. We’ve found that leading marketers are nearly 50% more likely than their mainstream peers to have increased their investments in capabilities like machine learning.4
When I started at Google Cloud in early 2016, I was happy to learn that we had a number of machine learning projects underway on our own team. One of them was for G Suite, our productivity and collaboration suite for businesses, which includes Gmail, Docs, Drive, and Calendar. Like many SaaS products, G Suite is promoted with a free trial period and we use online paid channels like display banners, search ads, and paid social to drive potential users to a sign-up page. In the past, we would launch a campaign with a mix of channels, wait for leads to complete the trial, then analyze to see what channels worked best, reallocate funds, and start the process over again. This process took several months, and we wasted money along the way on poorly performing channels.
So we asked ourselves, “what can we learn about the customers that have historically converted from trial to paid and how can that help us increase ROI going forward?” Working with data scientists and engineers, we combined data sets and used machine learning to analyze dozens of signals that suggested a customer was likely to convert. We created a model that helped us detect right away if we were attracting customers like the ones that had converted in the past. Within two days of advertising on a site, we could determine if we were advertising on a productive channel, and we could readjust spending right away. As new customers converted, the data model continued to learn—in fact, it’s still learning today. We are refining and repeating across markets and channels, at scale, freeing up dollars to invest in other marketing programs.
Where to start?
I’m often asked by other marketers about the best way to start using cloud technology to leverage data and build more customer-oriented marketing. I’ve found that it often works best for marketers to team up with a few data scientists, either internally or through an agency partner. When you have your first meeting, here are a few questions that could be good conversation starters:
- What are the top three data silos that could be combined to get a better view of our customer journey?
- What new customer insights could we unlock by combining our customer data?
- How can we connect audience insights to media activation and drive better performance?
While there are no quick answers to these questions, they can kick off discussions that will lead to entirely new ways of thinking about data-driven marketing.