Mike Fitzgerald is a Senior Analyst in Celent's insurance practice and an expert in property/casualty automation, operations management and insurance product development. Before he joined the research firm, he was vice president of enterprise underwriting Solutions at Zurich North America and his technology background includes the installation and maintenance of billing, policy administration and workers compensation systems.
Fitzgerald has been studying big data and analytics and recently helped Insurance Networking News judge corporate analytics implementations as part of INN’s VIP Award program.
He recently spoke with John McCormick, INN’s group editorial director, about meeting the challenges around big data. What follows is an edited transcript of their conversation.
What’s the biggest challenge companies are running into with big data implementations?
We conducted a survey -- it’s about a year old now -- of insurers and banks specifically around big data. For them, the number one area wasn’t technical and it wasn’t skills, it was actually cultural. Granted, there’s that word again. But we drilled down into it and found it has to do with a company being data driven and being committed to being data driven in its decisions.
Granted, it’s kind of odd when you think of all the actuaries and statisticians -- and the focus on numbers at these organizations. But what came back from this survey was that the culture of their companies wasn’t really data driven. So establishing a data driving culture is one of the first challenges.
What does data driven mean?
The distinction we were able to draw out was that data driven means you look at the data dispassionately and you make your decisions based on the data. And the feeling right now is that, in insurers and banks, there’s still a tendency to do it on intuition, or do it on legacy.
In contrast, data driven means you don’t rely on intuition to the degree that it has been. Examples of companies that are data driven include Apple, Google, Wal-Mart with their supply chain management; Toyota with its whole TQM [Total Quality Management] and lean process.
So it forms a pretty coherent story.
Now, interestingly, Progressive was the only bank or insurer that made the list. But if anybody was going to ask me which insurer would make it, I would have said Progressive.
Why?
What they did with automobile pricing.
In the innovation practice that I run, they are our example of a disrupter. And the reason that they saw the opportunity for disruption and then took advantage of it was just from the data and what the data was telling them.
However, in our survey, most companies said that their own firms were just moderately data driven. So that was identified as the biggest gap. Moderately is just not good enough.
What are the other big challenges?
That’s where we started getting into some of what you would typically see with an emerging technology -- lack of analytical talent, specifically around data science and quants, model building.
And another was just the pure lack of understanding about big data types of solutions. That shows there aren’t enough people around that really understand big data yet.
And when you talk about understanding big data solutions, are you talking about things like Hadoop?
Yes, the Hadoop platform and Ruby on Rails -- and how those technologies can assist with big data management.
And what about security? With big data pulling together so much information from so many different places, it seems like security would be a bigger issue than it appears to be.
And I think that it’s going to be even more important as we get more value out of big data.
So there are a number of significant challenges. What’s the best way for companies to overcome them? Where do they start?
Part of it is Leadership 101. And that has to do with communicating the growing importance of information. It has to do with really communicating to the company what they’re signing up for with big data.
There are companies that made a lot of money and are very successful and served their customers well for 100 years. They rely on their experiences.
But the world is changing very rapidly. We’re moving to being much more data driven. So there’s a need to push different messages and to push different approaches.
And the other is really investing in understanding of data -- what’s out there and what’s available. Understanding analytics. And then tying that back to the strategy of the company.
Organizations also need to invest in teaching people about data and how it lines up with their overall strategy.
For example, if you’re looking at a company and trying to learn about it, what can be gleaned from their data even if it’s not a public company? What information can you get from their websites, from blogs about that industry, from YouTube videos that could inform your knowledge? Those are the information sources that would be completely different than the traditional consumer data a company might collect.
This is a totally different way of thinking for many managers.
Most people running organizations, when they were cutting their teeth in the business, analytics wasn’t even on the table. You had reports. You had some business intelligence.
But you didn’t really have analytics.
So a lot of people leading organizations may be potentially missing a key skill -- a plain understanding of analytics and how it works.
Managers worked with statistics and financial modeling. But they didn’t actually sit down and do analytics exercises looking at, say, pies, basketballs, and oranges, things with no statistical correlation, and actually trying to see what’s related and what’s not related.
A lot of executives are talking about big data and the need for advanced analytics. But if leaders really don’t fully understand what big data and analytics are, then most organizations -- including insurance companies -- are really looking at an uphill climb.
The way I put it is that the people writing the checks for analytics don’t understand it. They understand policy admin. They understand claims. But this is something they don’t understand.
I’ve been trying to tell the vendor community, look, in addition to selling your tools, you really need to see what kind of education you can offer
But that’s tough. How are you going to tell the SVPs that they need to go back to school because they have a deficit in this area? This can be touchy stuff.
And is there a way to measure the progress of big data projects, to make sure you’re staying on track and making progress?
Well it’s not the traditional way. It’s not traditional ROI.
I think it’s very close to how you measure and track innovation. It’s more around how many different things you try -- how many attempts didn’t get past stage one, but also looking at how many got past stage two, how many got past stage three. It’s also how many insights did you gain that you never would have had unless you tried something to begin with.
So I would say the way to measure big data initiatives is the same why that you measure innovation, which is very different than the way that you measure typical projects.
