New applications of machine learning can help you not only react to your customers’ behavior but to learn new marketing strategies that actually shape that behavior. During the opening keynote of DMA’s 2018 Marketing Analytics Conference (MAC), the Chief Analytics Officer for Publishers Clearing House (PCH), Ash Dhupar, shared how his team uses analytics to optimize workflows across the organization.
For context, PCH is an interactive media company with a “tremendous amount of first-party data,” according to Dhupar. The company’s high audience engagement is driven by free-to-win sweepstakes and other games, and the company’s revenue is driven by omnichannel sales and programmatic ad solutions.
With over 110 million first-party customer profiles and 2 billion monthly page views on the PCH website, Dhupar says working for the company is “a data scientist’s dream.” PCH uses analytics to drive an improved customer experience across customer touch points, including paid media, emails, direct mail, apps and more. “On average we run about a thousand models on a daily basis.”
Lifetime Value Modeling
“The customer behavior hasn’t changed,” said Dhupar. “The way we interact with them or influence their decisions has changed” Dhupar said that, with all the focus on attribution and touchpoints, data scientists can lose focus on their purpose – the customer.
In order to maintain that customer focus, PCH built lifetime models, over 180 machine learning, non-linear algorithms, scored nightly, which predict the lifetime value of a customer to PCH. “If you start looking at what is the lifetime value that source is producing,” said Dhupar, “you can predict how much value each individual customer will likely drive for the company.”
The end result means that PCH is not focused people who are going to be one-timers or customer who will not enjoy the company’s experience. “Can I proactively get in front of this and get the right type of people?” Dhupar asked. “This is a win-win, driving revenue for the company and delivering a better customer experience.”
Transforming the Experience
Using these models in real time, PCH can determine whether a potential customer will be a high engagement or low engagement customer. The models also allow PCH to determine the customer’s propensity to buy.
This information was then used to improve the company’s email program. PCH sends 4.7 billion emails each year, approximately 90 million per week. “Sign up for PCH and you will see, way too many emails,” quipped Dhupar.
An Ensemble Optimized Customer Model allows PCH to determine a customer’s propensity to engage with emails, go dormant, or unsubscribe. This model scores 30 million customers on a nightly basis, and the marketing teams then use these scores to deliver customized experiences. This gave power to the business owners to scale back how many emails they were sending. This model cut down on the number of emails sent (by 200 million) and increased PCH revenue by $6 million.
With the right analytics foundation, a focus on profit and customer value, and an infusion of analytics, PCH was able to drive incredible innovation and an improved customer experience.
This week, hundreds of marketing analytics professionals are gathering in Atlanta for DMA’s 2018 Marketing Analytics Conference (MAC). Stay tuned to this blog or follow along on the hashtag #MAC2018 for more marketing and analytics expertise!