Guest Speak Part 3 (contd..)- AI in Claims – Workflow Management
Claims Director Brad Metzger has worked in the P&C Industry for several years, performing diverse roles in Claims Operations, Claims Strategy, and Claims IT. Having seen the rise of technology in claims management, he has a unique perspective in analyzing the potential of AI to significantly transform claims workflow management while validating it against emerging trends and risks. Below are his thoughts on how different stakeholders are impacted by AI and the takeaways on various aspects: how processes are eased, decisions simplified, and the role of analytics in bringing everything together.
Charlee.ai: Brad, what constitutes a claims workflow?
Brad Metzger: Claims workflow management involves steps that claimants and insurers must follow to resolve claims. These workflows are generally non-linear due to complexity and variability in each claim. That said, it does involve checkpoints in a pre-defined process to simplify the adjuster’s journey ensuring predictability for all stakeholders while bringing the customer back to pre-loss condition. It also ensures due diligence is carried out, considering their rights at every step. Insurers must fulfil their obligation to receive, investigate, and act upon a claim filed by the insured and take it to its logical conclusion as quickly and efficiently as possible.
In the real world, claims workflow management can be intricate, with various challenges and queries arising at every stage of the claims cycle. Frequently, insurers haven’t customized the process to map it to their deliverables despite following the universal pattern of how claims are dealt with. This can create chaos in the process;
- Essential data being missed during FNOL intake
- Documents not investigated despite being filed /received
- Specific claims not investigated deeply despite arising from anomalies identified in the past and /or unusual patterns and behaviours.
When ignored in the general management of things, these lapses can mean that significant insights are missing and can distort the way the claims investigation process flows.
Correctly mapping the claims workflow, understanding the complexity and variability that each claim will present, can resolve several critical areas of claims processing. Add to this the profound and predictive pattern detection by AI-led, NLP-based analytics, and the results derived by the application can change how claims are managed to make the process sharper, better, faster, and more accurate.
Charlee.ai: Brad, how can AI help ease and streamline claims workflows for the different stakeholders, such as Customers, Staff and Data Analysts including Data Engineers?
Brad Metzger: AI impacts everyone in the claims management process, helping align claims workflows for different stakeholders, injecting predictability, assisting with decision-making, and acting as a cyber supervisor.
Customer and Staff—AI eases communication when deployed at the customer’s first point of contact. It assists in automated claims submission (at FNOL) and ongoing bilateral
communication between the customer and staff using online portals, chatbots, or mobile apps, employing NLP to accept, understand, and process information.
It helps in the workflow management by;
- Ensuring less effort and more convenience for the customer.
- Reduces the need for human intervention and manual data entry.
- Eliminates the need for the claims personnel to complete mundane tasks, allowing them to focus on critical tasks required to adjudicate the claim.
Analysts – Data management is critical, but errors abound when done manually. AI helps efficiently manage large amounts of data to quickly analyze what’s impacting Service, Severity, and Productivity results. Analysts play a very crucial role in supporting the claims process. Acting as the CCO’s eyes and ears, they are responsible for addressing the factors that enables Adjusters to pay the right amount, at the right time, using the right amount of effort.
Charlee.ai: Brad, please share examples of specific parts of the claims workflow that are helped by the intervention of insights and past trends (For example, data extraction and document routing).
Brad Metzger: There are several critical milestones in processing a claim:
- FNOL – In any incident report and subsequent investigation, the basics must be addressed: Who? What? When? Why? Where? How? These are all of the essential first steps for the FNOL to be completed. Data must be accurately and efficiently registered into the claims management system, where AI’s intervention is very significant. AI’s pattern recognition software and ability to derive deep insights based on past claims patterns from structured and unstructured data are helping insurers predict claims trajectories in advance, allowing them to monitor the pulse of the allegations and the direction the claim is going. To know more on this, read the blog on FNOL Standardization.
- Damage Assessment—Vehicle damage after an accident is now being efficiently assessed using an AI-powered computer vision system. The assessment includes physical damage and component wear and tear.
- Fraud Detection: AI uses real-time data to identify and predict fraudulent practices, providing alerts during the claims workflow. It also uses image analytics to detect inconsistencies and anomalies indicative of fraud.
- Predictive Analytics—AI is increasingly used to analyze historical data to predict claim outcomes. Pre-trained NLP applications are seamlessly integrated with claims dashboards, providing actionable insights while predicting the trajectory of the claim into litigation.
- Customer Communication—Communication is one of AI’s foremost tasks naturally integrated into it. Chatbots increasingly used by insurers provide real-time updates and share information/documents, from the simplest asks to the most complex queries.
Document Processing – Sifting through reams of documents manually and then realizing some detail that could have changed the outcome of the claim is missed, is a
situation no insurer wants to be in. Here, AI can use OCR technology to extract information from documents, making it easier to process paperwork.
Charlee.ai: Brad, what factors in the claims workflow tend to drive poor claims outcomes?
Brad Metzger: While existing workflows are tried and tested, having come about from decades of learning, exploring, and practice, they have become repetitive and redundant and are slowing down the quick resolution of claims. In addition, several factors impact claims outcomes, such as the deluge of claims numbers, external volatility, economic upheavals, and other factors.
- Lengthy processing times – Delays frustrate customers who typically need to resolve the claim and get on with their life.
- Lack of communication – Lack of timely updates and understanding of the following steps can cause anxiety, leave people in the dark, and a build-up of resentment, causing the claims handling to become more complex.
- Inconsistent Service—Delivering the same kind of service inconsistently from one claims rep to another due to non-alignment with industry and company best practices, can cause customer discontent and escalate issues.
- Lack of empathy – A simple lack of understanding the other person’s point of view can lead to disagreement and dissatisfaction.
- Inaccurate assessment of damages—Several factors can lead to inaccuracy, one of the main reasons being hassled people who are overworked or overstressed and miss or ignore visible signs and texts in the claims data presented before them.
- Failing to follow up—Again, due to several factors, failure to follow up with the customer or any other provider responsible for servicing the customer (e.g., body shops, medical providers, etc.) can lead to miscommunication and missing crucial information that can impede or interfere with the claims management process.
Charlee.ai: How does optimal management of such workflows help insurers and policyholders? What are the takeaways or benefits?
Brad Metzger: Managing the claims workflows optimally begins from the FNOL when the claims are first reported, up to its resolution and various touchpoints. However, a claims workflow is progressing optimally when,
- It improves customer service and effort through quicker and more efficient settlements.
- Positively impacts paid severity (loss) and loss adjustment expenses (ULAE and ALAE). When workflows are redesigned and relooked at to minimize bottlenecks and streamline handoffs between different stages of the claims, the entire process ensures nothing is missed.
- Eliminating mundane, repetitive tasks encourages productivity gains (doing more work with fewer people) and allows people to manage more complex stages.
Charlee.ai: Brad, can you provide two examples from your experience of Charlee streamlining claims workflows?
Brad Metzger: In my experience with Charlee.ai, I have found NLP-based insights to significantly improve and manage claims portfolios. A single-pane view of all past and ongoing claims benefits everyone, gives updates on the general health of the claim portfolio, helps oversee reserve patterns, enhances the view of loss triangles, and provides many other proprietary key performance indicators (KPI) updates.
- AI lead analytics platforms enhance and optimize the “Standard Work” requirement (i.e., daily, weekly, and monthly tasks) for front-line supervisors. This improves focus on escalating claims involving high-cost or high-risk cases ‘very early’ in the life of a claim, akin to employing a Cyber-Supervisor or AI-led “virtual assistant.”
- Automatic triage of ‘all’ claims based on risk level without human intervention. The claims workflow process will be optimized when the right Claim Adjuster handles the proper claim, particularly on high-risk claims. This optimizes time utilization, hastens resolution, and prevents escalation leading to complex claims. It also eliminates the rerouting or reassigning of lower complexity claims.
Charlee.ai: Brad, how is the humungous amount of data generated from litigation perfectly aligned for AI to function and help grab insights from it?
Brad Metzger: AI is built to analyze large data sets and can quickly identify patterns and extract actionable and meaningful insights, resulting in ability to:
- Enhance and optimize decision-making processes for Litigation Adjusters and Legal Professionals when outlining claim-level strategies.
- Predict likely outcomes based on historical data.
- Detect inconsistencies and unusual patterns in data which is crucial for identifying potential direct or indirect fraud.
Charlee.ai: Brad, are there any surprising and unforeseen benefits of AI in streamlining and standardizing claims processes?
Brad Metzger: Yes! AI provides accuracy and efficiency gains in the claims process, which can enhance reserving best practices within the claims organization. This is achieved chiefly by leveraging advanced data processing to best predict outcomes.
A surprising outcome of applying Charlee to one of our claims processes was reserving accuracy. This was not a KPI, which we agreed on measures at the outset, but it ended up being a strategic insight and value-added consequence. Sharing the insights with the Enterprise Actuarial, Underwriting, and Pricing teams allowed them to enhance several existing risk models.
Written by: Brad Metzger
Claims Director
and
Dr. Charmaine Kenita
Technology and Business Writer
