Originally posted on LinkedIn, September 11, 2025;
They called it an upgrade. A new claims system with a shiny dashboard that promised better oversight, cleaner files, and fewer surprises. It tracked activity down to the minute: letters sent, diaries updated, closures logged. Compliance ticked up. Adjuster responsiveness improved.
Then six months later, an audit uncovered $3 million in preventable overpayments. The red flags had been sitting in claim notes and documents for months—data that the expensive claims system collected but never analyzed for quality in real time.
This is a scenario that repeats across the industry because claims technology was never designed to measure what matters most.
What current tools can do
In Part 1, we explored how critical errors hide in plain sight. In Part 2, we showed how detection delays compound the cost of mistakes. Now, in Part 3, we're going upstream to examine the flawed assumptions built into most claims software.
Modern systems are great at:
- Tracking procedural compliance with precision, logging when letters are sent, diaries updated, and files closed
- Maintaining documentation standards to support audits and regulatory requirements
- Monitoring performance by adjuster caseload, bottlenecks, and cycle times across claim types
- Providing operational visibility via dashboards showing metrics that were once invisible
For simple, low-severity claims with clear fact patterns, this works. But high complexity claims require critical judgment, which no system today can assess.
The Automation Blind Spot
To close the gap, some insurers bolt on rules engines and automated alerts. These tools help catch clear-cut procedural misses: missing documentation, late diary entries, or incomplete template fields. They validate inputs, not outcomes.
Take a commercial liability claim with disputed causation, overlapping policy years, and evolving evidence. A rules engine might confirm that deadlines were met and documents uploaded, but it won't show whether the defense made sense, the coverage stance aligned with case law, or reserves reflected real risk.
Automation keeps the trains moving. It never checks if they're on the right track. As a result, the most expensive mistakes—decisions that look fine until they derail everything—stay hidden until the damage is already underway.
This problem is magnified in complex, high-severity claims, where standard QA frameworks collapse under nuance. Designed for low-complexity files, most QA processes rely on procedural checklists that assume linearity: a step-by-step path from intake to resolution. However, high-severity claims rarely follow a script. Facts evolve. Coverage layers interact. Legal strategies shift mid-claim. The variables don't just increase; they interact. And when that happens, judgment, not procedure, determines whether the claim stays on track.
A poorly supported denial gets issued within hours instead of days. An inadequate reserve gets booked because the system confirms that all required fields are populated.
That's the reality in most operations today: high velocity, low visibility. Fast-moving, with no mid-course correction.
According to Bain (2024), generative AI has the potential to reduce loss-adjusting expense by up to 25% and claims leakage by 30–50%, but only if it's embedded into workflows that actively assess execution quality. Layering smart tools on top of old processes won't catch the high-severity errors that hide in plain sight.
Evaluating Judgment in Real Time
That's why leading claims teams are rethinking QA entirely—not as an after-the-fact review, but as a real-time, judgment-focused intervention model.
In our industry, quality assurance goes by many names: peer review, audit, oversight, quality control. At Elysian, we call it Dynamic Claim Review, or DCR, because it enables us to evaluate claims dynamically open or close.
The goal isn't just broader review, but also better visibility. DCR makes the thinking behind every file visible, assessing whether key decisions reflect sound, timely judgment based on the facts at hand.
That's the standard we apply across the board, including on claims we manage end-to-end. Every file, every decision, held to active scrutiny in real time.
As Elysian CEO Grace Hanson puts it:
"The fundamental design flaw in most claims technology is the assumption that task completion equals quality execution. You can have perfect documentation, perfect compliance metrics, and flawless workflow adherence while having bad decisions in a file that ends up costing millions. Ensuring quality outcomes requires evaluating whether strategy, timing, and evidence interpretation align with the facts of a claim, not whether adjusters checked all the boxes on a dashboard. That's why we built our approach around continuous judgment assessment. Because technology either amplifies good judgment or accelerates bad decisions at digital speed."
Takeaway
Digitizing a flawed QA model only helps bad processes run faster. The real lever is portfolio-wide, in-flight visibility into decision quality, so leaders can intervene before errors become final outcomes. In our next piece, we'll break down what it takes to measure judgment at scale without slowing the work.
This is the third installment in our series Delivering Claims Results. In Part 4, we'll explore how to unify fragmented oversight processes to eliminate the gaps where quality lapses persist. Don't forget to subscribe to The Elysian Era so you don't miss the next article. Follow Elysian on LinkedIn for more.
