Posted by Guest Blogger, Ian Turvill.
I just wrote an article on the use of Enterprise Decision Management to eliminate losses due to "claims leakage" in insurance. (See "EDM plugs the claims leakage dam".) It appeared in recent online and print editions of Fair Isaac's ViewPoints magazine.
From my point of view, the article does a good job of explaining how EDM can have a positive impact on the effectiveness of processing insurance claims.
This remains a very important issue for the industry. For example, a recent study of injury claims from auto accidents in Massachusetts found that 48% have some appearance of fraud or abuse. Yet, in spite of increased spending, three of five P&C insurers say their efforts to combat this problem are only moderately effective, or lower.
Unfortunately, because of space considerations in the online magazine, we had to cut out details of the so-called Claims Analytics Diagnostic. The Diagnostic is an approach insurers can use to identify where the biggest opportunities for improvement exist, and what the payback on their investment is likely to be. So, I thought I'd use the "unlimited" real estate that James affords me in this blog to add some additional detail about that approach here.
The Claims Analytics Diagnostic: The Reasons
You can’t manage what you don’t measure. So how do you measure and improve your organization’s ability to prevent fraud and other forms of claims leakage when – by definition – you don’t know what you’re missing?
An approach we recommend insurers deploy is the Claims Analytics Diagnostic. Insurers who apply this approach examine the outcomes of a sample of settled – or “closed” – claims. Staff should evaluate each claim against a number of leakage criteria, to determine the accuracy of prior decisions, and to find ways of improving decision making across all parts of the claims cycle.
For each prior claim among the sample drawn, the following questions are posed: “Was this claim processed correctly? Should the payment have been made? And should the payout have been smaller?” The answers to these questions indicate whether or not sources of otherwise avoidable leakage exist.
The Claims Analytics Diagnostic: The Theory
In effect, a Claims Analytics Diagnostic measures several dimensions of Decision Yield as they relate to claims processing. Specifically:
- Precision: Are the right claims decisions being made? For example, can you easily spot fraud, and are you referring the right claims for further review by the Special Investigation Unit?
- Consistency: Are they being made the same way across the organization?
- Agility: How quickly are new guidelines for claims handling incorporated into the claims administration process? For example, when new patterns of fraud are discovered, how long does it take for them to built into the claims review process?
When a Claims Analytics Diagnostic is completed, it provides invaluable guidance for prioritizing remedial investments. For example, should new systems be focused on addressing fraud first, or are there bigger opportunities worth pursuing in subrogation?
A Claims Analytics Diagnostic also provides the baseline from which subsequent improvements in claims handling can be evaluated.
The Claims Analytics Diagnostic: The Practice
While a Claims Analytics Diagnostic is relatively straightforward, completing one quickly and smoothly requires a good working knowledge of the steps involved and of the potential pitfalls. It also demands in-depth knowledge of your organization, your policies and sources of data. The diagnostic process has three major phases:
- The Preparation Phase establishes the scope of the exercise. A well-executed preparation phase is vital, because as the adage goes, “If you fail to prepare, you prepare to fail!” The main objective at this phase is to put in place the required tools, templates, guidelines, procedures and training. Also use this time to establish the processes needed to generate the data required for the review and to capture the data generated by the review itself.
- The Execution Phase requires careful management of the overall assignment of work across the claims professionals involved. During this phase, a carefully designed sample of claims is drawn, and the outcome of each claim is revisited and evaluated. The team should focus on securing the necessary total volume within the time available, without compromising the quality of the analysis.
- The success of the Analysis Phase depends on the care and attention applied to the prior phases. It involves basic housekeeping steps, such as data cleansing and reconciliation across the different analysts involved. However, it is primarily about finding and sizing the opportunities to reduce claims leakage. Typically the team presents three kinds of analysis:
- Case studies, which provide tangible examples of how existing claims decisions may be broken
- Leakage spreadsheets, which provide line-by-line descriptions of the sources of leakage
- Bespoke analysis, which responds to the team’s specific areas of interest and lines of questions
The Claims Analytics Diagnostic: The Results
The end result of this process is a clear understanding of how much you’re losing to various forms of claims leakage, as well as how and why they are occurring. An important further outcome is a set of recommended process and system improvements to reduce or even eliminate these claims losses.
The recommendations will have a particular, though not exclusive, focus on improvements that are possible through the application of Enterprise Decision Management (EDM). As readers of this blog are well aware, EDM is a systematic approach to automating and improving decisions across the enterprise by embedding predictive analytics and decision automation technology into transaction processing systems.
The use of EDM to improve claims could include, for example, the use of statistical models to identify the likelihood of fraud given the information that is made available at the First Notice of Loss.
A representation of the high-level results of a Claims Analytics Diagnostic and the associated EDM-related recommendations is shown in the attached diagram. See how the underlying causes of "avoidable sources of claims loss" have been identified and quantified, and how each root problem is addressed by a different approach based on predictive analytics and decision automation. This top-down view is invaluable to senior insurance executives who are seeking to rationalize and prioritize their investments in claims improvements.
To see the most up-to-date index of ViewPoints articles online, click here. If you want to subscribe to ViewPoints, which like our blog deals in-depth with EDM, click here. Caveat: Given that ViewPoints is a Fair Isaac corporate publication, it tends to be much more salesy in nature than this blog.