Over the last decade, data and analytics have grown to be more than just a quantitative support function. Many organizations have traditionally used data to win customers and market share. However they are now also leveraging data to re-design future products based on evolving customer needs and macro trends. While significant progress has been made in the field of machine learning, as well as artificial intelligence –there is one critical element to making this all work: having the right data. Business decisions that are built using flawed data can cause an organization significant revenue loss, increased expenses, compliance issues, possible legal issues and even more severe ramifications.
The focus on machine learning is making data even more essential than ever. At the recent Data for AI 2020 conference, Shiv Misra who is the Head of Medicare Retention Analytics at CVS Health shared how he compares data to oxygen, how companies can support actionable insights from their data, and examples of how data has successfully been used throughout his diverse and impressive previous positions at Fortune 500 organizations. In this article, he further shares insights on these topics.
Why is it so important for organizations to think about their data – from a collection, usage, and privacy perspective?
Shiv Misra: Throughout my experience within data science and analytics, I’ve noticed a general tendency amongst businesses to glorify advanced analytics while treating data simply as the necessary first step. I have also seen a lot of sophisticated models and insights that could have driven large business impact, being tossed out of the window because the data that was used missed the mark on either accuracy, precision or timeliness. I therefore strongly believe that having an organization’s data accurately connected across all internal departments and with 2nd+3rd party sources can truly power winning strategies.
How do companies support actionable insights from their data backed by advanced analytics?
Shiv Misra: The honest truth is that while data science and analytics can help mine incredibly powerful insights, it takes strong partnership with business to drive real change and a sizable impact. This partnership with business is founded on two critical pillars i.e. Trust and Accountability.
To highlight the use case for Trust, I would love to share an example from my role at PayPal where I was tasked by the leadership to grow the usage of decision science solutions within the organization. I was surprised as to why will that be needed at an organization that is built around data. After a series of surveys, interviews and 1×1 discussions, I realized that since there were thousands of data savvy decision makers within the company who had access to hundreds of data sources. There was an overall lack of trust in the data that was being leveraged by the analytics team and hence not many users wanted to log-in and utilize the powerful decision science engine that our team had created. We found a novel fix for this challenge, which was to ingest the ‘trusted’ data source into our engine and show the variance in our data vs. the trusted source of truth, using a 5-star rating scale. This allowed for business partners to truly know if the data was timely but only directionally accurate (for Marketers to use) or was extremely close to the system of record (for Finance teams to use). Trust in our data drove the usage of our solutions significantly.
Data also forms the foundation for accountability, and I’d love to share my experience as someone who started and built the data science and analytics function at Heineken USA. My team was tasked with building the first of its kind commercial effectiveness model for the CEO and that involved leveraging data from almost all departments within the company. We knew that our insights would drive significant transformation in ‘ways of working’ across all key departments. For this very reason, we started our work by spending a great amount of time in building a solid data harmonization engine that aimed at bringing disparate data points in perfect unison with each other. In the process, we made sure that our business partners were not only consumers of the insights but also became true champions of our approach. Data harmonization and a series of trainings helped us make our stakeholders aware of the data sets that their organization owned thereby driving up their ownership in the process. This helped us build and launch the first of its kind optimization engine that drove significant incremental value for the business.
You like to think of an accurate data set as the ‘Oxygen’ for any good statistical analysis. Why do you call data the ‘Oxygen’?
Shiv Misra: I feel that any advanced analytics model or approach is futile if it is built using data that is not accurate. Even when the fuel runs low the fire still burns albeit a bit dim, although when oxygen is cut off the fire dies. Similarly, incorrect data can cause the entire data science models and insights to lose value. Hence I prefer calling data as the oxygen and not the fuel for the decision science engine.
What is Data harmonization?
Shiv Misra: With better data tracking capabilities and a higher awareness for analytics, organizations now have access to multiple data sources with large volumes of very detailed information. While data availability is not much of an issue today, it is the ability to make sense of these large data sets that can give organizations a competitive edge. Data harmonization is the process of combining diverse data sets in such a way that the format, dimensions (like time period, geography etc.), nomenclature and definitions start following the same tune. This cohesive data set could power robust data science and machine learning programs across all levels within an organization.
What are some of the insights and opportunities for dealing with data and AI from a healthcare and insurance perspectives?
Shiv Misra: The potential for Artificial Intelligence within Healthcare and Insurance is immense. Studies have shown that AI can augment the human intellect when it comes to diagnosing a disease faster and in a more accurate way. Aetna is using data and advanced analytics to drive early intervention which can keep people healthier while reducing medical costs. This was one of the reasons that motivated me to join Aetna which is part of CVS Health, an incredible Fortune 5 company that is ‘Helping people on their path to better health.’ More details could be found here:
What are some of the challenges for dealing with data and AI from healthcare and insurance perspectives?
Shiv Misra: I am too new to list all of the challenges with data and AI within Healthcare but based on my past experience across multiple industries, I can say that most challenges are industry agnostic. I have touched upon these in my response to the earlier questions, but to summarize here are three key areas that I feel are important to address correctly:
- Building trust within business partners to showcase that AI models do work
- Driving change in the ways of working, esp. at large organizations
- Recruiting the right talent with data skills but more importantly the passion to drive transformation
What do you see as critical needs for workforce development around AI?
Shiv Misra: While we have plenty of data science and machine learning programs that have started building large pool of fresh talent within analytics, there are a few soft skills that I believe help build winning data science teams:
- Integrity: The most important skill for anyone working within AI is to be honest and custodian of truth
- Passion: It is important for the workforce to enjoy the process as much as they cherish the end result, esp. within AI and Machine Learning
- Curiosity: The mindset to keep finding ways for improvising is the foundation on which AI was build. Curious data scientists are often the ones that learn the most
- Adaptability: Pace of change within AI is enormous, and hence being adaptive is synonymous with being the winner in this space
- Resourcefulness: Not always do we have the best tools or 100% complete data, it is critical to be scrappy and have the mindset of a hacker
- Communication: As discussed earlier, the key to success of an AI program is to drive adoption within business through a robust communication mechanism
What AI technologies are you most looking forward to in the coming years?
Shiv Misra: I believe the future belongs to cognitive analytics. With advanced computing capacity that is increasingly becoming available, large data sets can now be perfectly combined and advanced AI models may soon allow for machines to make human life substantially better. This includes machines making most if not all of the basic decisions for us, freeing up the human intellect to find ways to speed up growth. Augmented Intelligence will allow for mankind to drive transformation at a pace that has never been seen before, and I am incredibly excited to be a small part of this big evolution.