Automating customer engagement with big data analytics

For years, companies across the retail, telecom, insurance, and banking industries have used machine-learning techniques to analyze terabytes-upon-terabytes of real-time data to improve customer engagement.

This kind of closed-loop data analysis pulls from an extensive palette of customer interactions and lifestyle events to paint increasingly accurate portraits of consumer behaviors. Naturally, the ability to predict future actions and events results in improved engagement. And that can deliver significant value to organizations.

But this type of large-scale analytics has been underutilized by players in the American healthcare industry. However, this is beginning to change. Especially for payers, who are increasingly determined to lower the total medical costs for their member populations within the new value-driven care paradigm.

At EY, we are currently working to develop at-scale data analytics programs through the utilization of real-time data seen in other industries.

As the first step in this process, we ask two key questions:

1. How can you better anticipate patient needs along their journey to drive better outcomes?
2. How can you quickly identify an event and take the best action to help the patient in that moment?

To address this, EY introduced a “layered” data platform to integrate not just claims and demographic data, but also the additional inputs utilized by other industries such as call center, web interaction, and mobile phone interaction data. We call this platform a “time- sequenced customer journey data lake.” It helps payers anticipate future customer behaviors, signal deviations from expected behaviors, and even prescribe new interventions.

With this approach, we’ve found any proactive outreach to be inherently positive, but it is the timing of that outreach during patient journeys that’s critical to patient/member engagement. For example, through a recent analysis of one of EY’s clients, we saw there was a negligible 4% increase in customer engagement if they conducted outreach before any key journey steps, such as a specialist visit or pharmacy claim. However, they saw customer engagement skyrocket to 83% when they conducted outreach following one of the key journey steps.

A layered journey analysis also allows us to prioritize customers based on their “likelihood to engage” and automatically inform us when their likelihood is the highest (timing) so we can conduct outreach at the most optimal moment. Likelihood-to-engage profiles have been utilized by others in the past, but we have incorporated the time-sequenced analysis of the specific journey steps and their correlation to likely engagement.

With EY’s client, we applied a time-sequenced data analysis to 43,000 patient journeys specific to oncology. The results showed that if they were to use this approach with the top 10% of oncology patients in the program, they could improve engagement by a staggering 159%. Assuming they could apply this to all their customers, we found that the time-sequenced approach could result in a 30% increase in engagement overall.

Needless to say, these and other findings spell out massive value capture for organizations willing to implement it. But the rising adoption of machine learning and journey analysis among payers is an indication of something much bigger: A consumer-centric mindset is beginning to take hold in the health industry.

Integrating this approach will certainly not be without challenges for those unaccustomed to it. Luckily for them, however, other industries have already devoted the blood, sweat, and tears to build these layered data platforms.

The successful health organizations of tomorrow will be the ones who chose to paint from the same data analytics palette.