To succeed with data-driven decision making, insurers and data scientists must accept that how one learns is more important that what one knows, and that humility and change management are critical for establishing influence.
AIG Property and Casualty has been a pioneer in data science. And, Murli Buluswar, AIG’s chief sciece officer, took the opportunity at ACORD LOMA in Orlado this week to share his observations and insights on the capabilities and opportunities offered by this burgeoning technology with attendees, and Tom Koulopoulos, president and founder of Delphi Group, who hosted the interview.
“When the rate of change outside your organization outpaces the change within the organization, the end is near,” Buluswar said, quoting Jack Welch, former CEO of GE. AIG has accelerated the rate of change within the company by aggressively embracing data-driven decision making; the global insurer now has more than 125 people on the data science team, Buluswar said. When hiring, grade point average is not the number one criteria, he said. Rather, Buluswar says he looks for intellectual curiosity as evidenced by based decision making, creativity and resilience, ability to learn and teach, humility to know its about how you learn and adapt, requires new thinking and new approach.
“It isn’t about what you know, it’s about how you learn,” Buluswar said. “The vision behind the creation of the science team was: As the world around us changes exponentially, the winners will win big, but they will be removed from their perches faster. So it requires a fair bit of agility.”
See also:
Agility, in the context of the discussion relates to not only the creation of new problem solving abilities, but the requirement that those abilities are then brought to bear on the actual problems insurers are confronted with.
“The thinking that created the problem can’t be the same as for the solutions, Buluswar said, paraphrasing Albert Einstein. Data driven decision making, he said, can be simultaneously incremental, transformative and disruptive. Buluswar acknowledged that some people resist learning from data and don’t always know what questions they should be asking, which can lead to them solving for symptoms and not the actual cause of the problem.
For example, if a line of business is losing money, simply raising rates can result in the loss of both good and bad business. “You chase away your better customers at a faster rate, you’ve shrunk your top line and now you are worse off. Thinking about that context and recognizing the fact that you might have a certain view about what the wanted exposure is.
“People can examine the same risk and come to radically different conclusions. It’s an opportunity to learn from the collective wisdom of what the rest of the industry is doing.”
The promise of data science, Buluswar said, is that it can help insurers to not just measure risks better, but actually help lower risks, creating better outcomes for both customers and insurers.
An exchange about life insurance, and the fact that 40 percent of applicants in generations Y and Z drop out in the application process due to its intrusive and time consuming processes, prompted insights from Buluswar about the opportunities for insurers to embrace technology in order to help more people take advantage of insurance products.
For example, genomics, which includes molecular biology the use of a person’s DNA and RNA to understand individual health risks, combined with data science, could transform health insurance. But, while genomics is enormously promising from an underwriting standpoint, privacy and other concerns could slow adoption, which also is the case for insurance telematics.
As consumers, Buluswar said, we are voluntarily and happily emitting enormous streams of data. “That creates potential threats in terms of how public and private enterprise take advantage of that data and what they do with it. On the other hand, it’s becoming a part of our lives. We are comfortable emitting all sorts of data every single day about everything we do. And we do so voluntarily quite often,” Buluswar said. “Whether it is genomics or telematics, we will embrace it because it expands the marketplace. It gives more people more access to insurance, who otherwise wouldn’t be able to, and it influences the outcome. We are not just measuring the outcome. Telematics, in some parts of the world, is already changing driving behaviors because it is creating a feed back loop. That’s a beautiful thing and that’s the direction the world is evolving.”
The best data scientists, Buluswar said, understand this difference between outcomes and output, and are change leaders, influencing the culture in terms of how decisions are made and what decisions are made. However, success requires not only speaking to logic, but also having the humility to understand what they can and cannot yet know.
While some CIOs may feel threatened, marginalized and commoditized by data scientists, Buluswar offered the consolation that in this technology-driven time, CIOs are the business. Data scientists are responsible, on the other hand, for bringing a discipline, that being data-based decision making, to the organization. However, Buluswar added that CIOs’ sense of fear could be a good thing, if it induced them to be more reflective about whether they are holding their employers back.
Change management and empathy are important attributes of successful data scientists, Buluswar said, because while data driven decision making is powerful, it can be ineffective if it doesn’t acknowledge that people are emotional, and that fear and greed are powerful motivators. “Focusing just on the numbers leads to incomplete answers,” Buluswar said.
Change will be difficult, he offered and data scientists should pick areas for wins, make mistakes quickly, learn from them and correct course. ”Fail more intelligently,” Buluswar said, and don’t consider failures as permanent. ”Buy political capital with the courage of conviction,” he said.
One mistake Buluswar said he had made initially was hiring mid-career professionals, who lacked the gravitas to influence senior staff members. To be influential, Buluswar said, the scientists needed to be able to guide others, and senior staff most importantly, into a world of possibilities. And, realizing that lack of a good fit, Buluswar said the solution is to act quickly, offer a cushion to the outgoing professional and act with trust and transparency.
In closing, Buluswar said leaders can increase innovation by reducing the risks associated with failure. Belief in one’s self is always important, he said, because the odds are always stacked against us; he also cautioned against becoming a slave to pessimism. “If you are going to drive change, its’ not all in the numbers,” Buluswar said.
