We hear a lot of talk these days about the virtues of data-driven organizations. That’s certainly reasonable up to a point — but what does it really mean? When it comes to routine operational decisions, in particular, the current bias seems to favor increased automation over human judgment. The data doesn’t lie — or so the story goes — so we’re better off deferring to programmatic decision models.
That notion may be reasonable for some situations, but when you’re operating in a complex and nuanced domain like casualty insurance claims, that highly automated decision paradigm can begin to fall apart very quickly. Thousands of different variables come into play. Medical records and accident reports contain subtle details that provide vital clues about potential risks. To complicate matters, important minutiae are often buried deep inside the narrative content.
An experienced claims manager can pick up on that nuance, provided they have adequate time and attention to devote to reviewing the documentation. Can an algorithm accomplish the same thing?
The short answer is yes, but that comes with a vitally important caveat. In complex domains, advanced data analysis should not drive automated decisions; it should inform and empower human beings to make more effective decisions. The most effective artificial intelligence (AI) initiatives in place today are doing exactly that.
Data-Driven vs. Data-Informed
The distinction here is critically important. The data-driven paradigm is about automation. It’s about shifting decision-making responsibility away from human actors and trusting the algorithms to take their place.
A data-informed approach, in contrast, empowers and assists people to make better decisions by flagging potential risks, highlighting anomalies, and monitoring for changes that may indicate a need for attention. It’s a helper, not a replacement.
For claims managers, this approach has powerful implications. Imagine, for example, that an injured worker has missed three consecutive appointments for physical therapy. What does that mean? If the employee no longer feels a need for treatment, then it may be a sign that they’re ready to return to work, but it could also be an indication that the case has taken a turn for the worse. In either case, an adjuster should be made aware of the situation so they can make a proper assessment.
In a complex domain like claims management, this data-informed approach holds tremendous potential for transforming organizational culture and processes.
Consider how claims are handled today at most organizations; adjusters typically follow up with cases on an “as needed” basis. Depending on the individual adjuster, that might involve a diary notation, a running to-do list, or a collection of sticky notes. Inevitably, though, it means manually reviewing medical bills and records as they come in or when the adjuster’s schedule permits.
In a data-informed organization, claims adjusters focus on meaningful decisions. Because they no longer need to spend their time scanning records in search of salient information, they have sufficient bandwidth available in which to apply their professional judgment on high-priority cases. AI does that legwork for them.
Data-informed organizations can apply their valuable resources toward predictive severity-based workloads. They can focus on claims that need attention today — based on real-time data. Incoming documents are reviewed and scanned by AI, and claims adjusters are notified when a case requires their attention.
See also: The Data Journey Into the New Normal
The Business Opportunity for Insurers
The data-informed approach is already operational in a number of leading companies around the world. It’s transforming processes and driving cultural change — but not in the way that many AI skeptics have predicted. Data-informed organizations aren’t dehumanizing their processes. On the contrary, they’re empowering and elevating their claims professionals by enabling them to focus on meaningful work.
The data-informed paradigm is about focusing on the right claims at the right time. It’s about spotting correlations and anomalies, identifying potential risks and bringing those to the attention of an experienced claims manager.
The result? A data-informed organization has a shorter claim duration and lower-than-average total claim costs. Not surprisingly, workers at data-informed organizations also enjoy substantially higher job satisfaction. These companies are generating high ROI — not by reducing their workforces but by elevating them to higher-value activities.
The Build vs. Buy Debate
How does an organization achieve that kind of transformation? It starts with a predisposition toward innovation and a recognition that advanced data analytics has the potential to transform claims management from an operational perspective.
Conventional wisdom tells us that proprietary data is a differentiated asset. In other words, companies place a high value on their internal data because it’s theirs, and nobody else has it. In the world of AI and machine learning, though, more data is generally better. When ML models have access to higher volumes of information, from a relatively wide array of sources, they can “learn” faster and more effectively.
Building and maintaining those kinds of high-volume data sets can be extraordinarily costly and time-consuming. The implication for insurers is that, in the build-versus-buy debate, there is an increasingly powerful case for moving beyond proprietary data and embracing best-in-class platforms to drive the data-informed model.
This provides for a flexible co-innovation process, enabling insurers to leverage solutions and platforms that have already been proven in the real world, without reinventing the wheel. It’s the fast-track alternative for companies seeking to become data-informed organizations.
As first published in Claims Journal.
