Special Investigation Units (SIU) are extremely process-oriented and follow well-documented procedures to decide when a claim should be referred for investigation and what actions should be taken. Most of the staff are seasoned investigators who may be more inclined to trust their experience and tried-and-true processes than analytical techniques that may seem unintelligible.
They can’t trust the data if they don’t understand it - raw analytical model output is often full of technical jargon that’s not easily understood by the business user. This creates frustration for the end user - often an analyst or investigator with more business expertise than technical or statistical experience – and may lead them to think that the analyses are meaningless and analytics lack any real business value. To convince these users to adopt analytics, the solution must give them easily consumable information.
And, you must integrate the analysis with the business process. Fraud analytics solutions need to meet the investigators’ business requirements and be structured in a way that makes sense to them – not just to the data scientists and IT teams. They need a well-configured user interface (UI) that gives them all the information they need to make a decision about whether or not to proceed with an investigation.
Here are five things you can provide to help them see analytics as the answer:
- Provide a “one-stop-shopping experience.” The investigator should be able to complete the triage and review process without ever leaving the UI. In a manual fraud detection environment, the investigator has to access multiple internal and external data sources. By providing instant access and a holistic view of these data sources, you exponentially increase investigator efficiency and start to win converts.
- Remove the technical jargon. Use scorecards that speak the business language to show the investigator the rules or scenarios that were surfaced by the model and how much weight each contributes to the overall fraud propensity score. This will help them understand why a referral scored high for fraud risk and allow them to focus their investigation on the most critical factors.
- Show and tell. Analytical output surfaced in the scorecard should be supported by other data in the UI. Incorporate sections or links to reports that contain the data that support the rules and scenarios triggered in the scorecard. Investigators will not – and should not – blindly accept that the fraud scenarios surfaced in the scorecard are factual. Providing easy access to this data also helps facilitate their investigations.
- Leverage third-party data sources. Many SIU’s invest in expensive external data sources and most aren’t integrated with internal data sources. The lack of integration makes it difficult to get the most value from the data. Insurers can get more value from their investment by incorporating it into the UI, combining it with other data elements, and displaying it in a meaningful and user-friendly format. Common examples of this third-party data could include medical bill audit data, industry watch lists, sanction lists, loss history data, estimate data and public records information.
- Give them more than just a score. Provide deeper insight into suspicious activity by incorporating data visualization techniques such as link analysis, maps, graphs, charts, annotations, and highlighting of key text and other data elements that support the scorecard findings. These simple approaches resonate with the end user and add tremendous value by directing attention to the most relevant information.
The most powerful analytics in the world have little value if they can’t be readily understood and adopted by business users. Start with these five tips to get your investigators excited about fraud analytics. These will help them see that analytics can drive better decisions and better outcomes.
What do you think the biggest challenge is to the adoption of analytics? Tell us in the comments.