Reality is tough to capture. It keeps moving. But somehow we’re growing faster and better at capturing it.
Consider visual reality. In 200 years, we’ve moved from illustrations and paintings, through still photography and into motion pictures. We then created technologies to transport the motion pictures across space to the places we wanted it. We’re now looking at 4K televisions and talking to family with FaceTime or Skype on displays that have the same or greater resolution than our eyes.
Data’s reality is no different. Back in the late 1980s, I did work for a paint manufacturer, trying to monitor the real-time operating conditions in one of its paint plants. We connected some PCs to the plant’s programmable logic controllers and then asked the controllers every 30 seconds, “How are things going? What are you working on?” The controllers spit out lots of data on operating conditions. We charted, we graphed (in real time!), and the plant operators had new insights on how things were going with paint production.
We were augmenting the physical instrumentation of the plant with virtual instrumentation.
Instrumentation — Data’s Virtual Reality
So how is your insurance company instrumented? Are things running a little hot? Do you find yourself running short on any raw materials? How full is the pipeline? When do you find that out? Is it tucked into a spreadsheet a few weeks after the end of the month? Could you make more money if you found out in five minutes instead of five weeks?
Are “modern” insurers still living on static pictures of data’s reality?
Insurance leaders are creating real-time instrumentation for their companies, allowing them to open and close everything from granular geographies to wind risk and monitor premium production compared with last week, last month, last quarter, last year, as of today or any day.
To better instrument our companies we need to think about: acquisition and transportation; accuracy; presentation timing and type; automation and cognitive capabilities; and actions and reactions. When you finish this post, I think you’ll agree with me that instrumentation should carry a high priority in insurance’s digital agenda.
See also: How Virtual Reality Reimagines Data
Acquisition and Transportation of Data
How do we monitor the data in a flow of information in constant motion, not just the discrete sets that are static and in place? First, our goal is to NOT be another weigh station in a step-by-step process. We need to be tapped into the flow without impeding it. To do this, we set up measurement devices that allow us to peek into the flow and draw of our information, then shuttle it to where we need it. This is not unlike the earliest “vampire” network connectors, feeding on Ethernet cables as opposed to a light socket sitting within a circuit.
There are any number of tools that one can use for real-time streaming and visualization, but the key to having any of them working properly is the setup of the data acquisition. A vampire approach will allow for real-time monitoring, as opposed to relying on continual requests and responses from data sources.
Accuracy of Data
One of the challenges in looking at continuous data is that spurious results may throw off the averages, so we need to be careful about outlier events. When looking at real-time data, it is far more likely that outliers will appear.
For example, as I was driving the other day, one of the “Your speed is...” signs I passed registered 110 mph. (I’ve driven 110 mph before, but not this day.) It quickly corrected itself to 55 mph. Data “in flight” like that needs the right periodicity to make sure that it is capturing the 55s, not the spurious 110s. And data obviously needs to be trained on what to notice and what to ignore. Automated removal of outliers helps keep the data pure. Keeping a concrete set of rules regarding data’s use will be very important in allowing people to trust the data when it is presented.
Presentation Timing and Type
In 2007 and 2008, Starbucks began opening stores as an undisciplined growth strategy. Eighteen months later, many of them were shuttered in a massive restructuring. In 2011 and 2012, Starbucks was adding stores again, but this time based on GIS traffic-flow data and demographics. Real-time reporting had become a more valued part of the business structure. Former Starbucks CEO Howard Schultz reportedly received store performance numbers as frequently as four times each day.
How often an insurer needs data and how it wishes to have information presented is a matter of need and preference, but it can clearly be tied to business strategy. For one client we worked with, they realized that continual data visualization in public locations, such as lobbies and meeting areas, helped the whole community see how important data was to the decision process. Others may wish to keep their data tucked out of sight but still available via tablet or cell phone.
Depending on the insurer and the insurer’s reactive capability, they may want feedback every day, every hour or every few minutes. Whether you choose to use dashboards, standard reports or e-mailed updates will also depend on your role and your need to know.
Automation and Cognitive Use
One of the drawbacks to data visuals of any kind is that they are subject to perspective. Trends and movements can be hard to spot over time. Anyone familiar with Excel line graphs will understand what I’m talking about.
The graph below looks fairly flat. But it shows a 5% move from start to finish. Identifying that size movement will be important.
Here is where automation in data’s motion pictures plays an important role. If the system can “learn” what good performance looks like, then it can also improve its ability to communicate vital information in a timely manner. I was just on a call where we discussed facial recognition in insurance. The use case was that there are teams working to identify faces and emotions on faces. If we have tools that can tell if someone is unhappy, surely we can use those tools to recognize a hidden pattern in our data. Data’s flow won’t just represent current trends, it will also identify oft-hidden patterns. What we think we know from our common snapshot approach to data may be overturned when cognitive capabilities start to bring new insights to our eyes. Once again, data’s motion pictures aren’t just for our own amusement, but they greatly enhance our strategies and decisions.
Actions and Reactions
If I run a chemical plant, I’m deeply concerned with monitoring real-time flow. Every action I take to tune that plant has a reaction. As insurers, we should also be concerned with real-time flow, capturing our understanding of reality.
But there is also a historical component to data’s adjustments. In the chemical plant, if I change the mixture of a certain compound based on my data and the new mixture works, then I need to capture that moment in time as well. It is equally important for insurers to capture the timing of their corrective actions to make sure that we can see the relationship between action and reaction.
See also: Your Data Strategies: #Same or #Goals?
Overlaying notes to explain that “we reduced available capacity in less profitable zip codes in June” should show some point of inflection in our results. Having that as a part of our reporting is critical to creating the positive action, a reaction cycle that we want to reinforce.
We have an embarrassment of riches when it comes to data, and we are only going to get richer in the coming years. By instrumenting our organizations and realizing that we need some new tools and techniques to turn that information into actions that create the right reactions in our organizations, we can improve our results every day, week and month — not just when we close the books.
Here is where automation in data’s motion pictures plays an important role. If the system can “learn” what good performance looks like, then it can also improve its ability to communicate vital information in a timely manner. I was just on a call where we discussed facial recognition in insurance. The use case was that there are teams working to identify faces and emotions on faces. If we have tools that can tell if someone is unhappy, surely we can use those tools to recognize a hidden pattern in our data. Data’s flow won’t just represent current trends, it will also identify oft-hidden patterns. What we think we know from our common snapshot approach to data may be overturned when cognitive capabilities start to bring new insights to our eyes. Once again, data’s motion pictures aren’t just for our own amusement, but they greatly enhance our strategies and decisions.
Actions and Reactions
If I run a chemical plant, I’m deeply concerned with monitoring real-time flow. Every action I take to tune that plant has a reaction. As insurers, we should also be concerned with real-time flow, capturing our understanding of reality.
But there is also a historical component to data’s adjustments. In the chemical plant, if I change the mixture of a certain compound based on my data and the new mixture works, then I need to capture that moment in time as well. It is equally important for insurers to capture the timing of their corrective actions to make sure that we can see the relationship between action and reaction.
See also: Your Data Strategies: #Same or #Goals?
Overlaying notes to explain that “we reduced available capacity in less profitable zip codes in June” should show some point of inflection in our results. Having that as a part of our reporting is critical to creating the positive action, a reaction cycle that we want to reinforce.
We have an embarrassment of riches when it comes to data, and we are only going to get richer in the coming years. By instrumenting our organizations and realizing that we need some new tools and techniques to turn that information into actions that create the right reactions in our organizations, we can improve our results every day, week and month — not just when we close the books.
