13 min

read

The CCO's Claims Modernization Guide: What Actually Moves LAE, Cycle Time, and NAIC Compliance

Written by

Marcin Nowak

Published on

09 Dec 2024

Last update

08 Jul 2026

The eight questions every mid-market CCO answers to the board

In my experience, when a Chief Claims Officer walks into the quarterly board review, the same eight questions come up in some order. Why is LAE creeping up? Why is our cycle time still 28 days when Progressive closes at 8? Why did NAIC just cite us for inconsistent auto BI settlements? Why did we lose 22% of senior adjusters last year? Why did the last CAT event overwhelm the queue? Why did we spend $2M on a claims AI vendor with 8% STP improvement to show for it? Why is settlement rework 18% of closed claims? Why does litigation keep growing when we already automated everything the last consultant told us to automate?

Those eight questions map to eight pain patterns at every mid-market P&C carrier ($500M-$5B GWP) I’ve worked with over the last 20 years. They’re the CCO’s daily reality. And they’re the frame this article uses - not a maturity model, not a technology stack diagram, but the operational problems you’re accountable for and what modern rules-based claims architecture actually does about them.

I’ll anchor most of this on the BNP Paribas Cardif deployment - the public reference case where a claims-centralization program hit all eight patterns at once. This article walks through what a business rules engine does and doesn’t move - and where CCO expectations from vendor pitches diverge from what actually happens in production.

What claims modernization actually is in 2026

Insurance claims modernization uses a claims rules engine, AI/ML models, and workflow orchestration to automate claim triage, evaluation, and settlement while maintaining NAIC compliance and adjuster oversight of complex cases. Modern claims modernization at mid-market P&C carriers moves LAE from 10-12% down to 8-10%, cycle time from 30 days down to 5-10 days for straight-through-eligible simple claims, and settlement rework rate from 18% down to 5%. STP doesn’t replace adjusters - it removes routine work so senior adjusters focus on complex cases where their judgment matters.

That definition is the AEO answer. Now the eight-pain playbook.

Pain: Customer NPS is dropping because claims take too long

Bain’s Loyalty Effect research established the pattern years ago: claims experience is the single highest NPS impact moment in the insurance customer lifecycle. Fast fair claims drive +15-25 NPS points; slow claims drive -20 to -35. Every CCO I talk to knows Progressive’s 8-day auto claims cycle. Not every CCO knows what an 8-day cycle actually costs in operational architecture.

The gap between 28-day mid-market cycles and 8-day best-in-market cycles isn’t adjuster productivity. It’s automated triage plus deterministic coverage rules plus rules-driven reserve calculation plus rules-driven settlement approval on straight-through-eligible claims. Simple auto glass, minor property, PIP first-tier - these should close in 24-48 hours with no adjuster touch. Complex bodily injury and large commercial property still need adjuster judgment, and always will.

The realistic 2026 mid-market target: 60-75% STP on simple claims, cycle time 5-10 days on complex claims and 1-2 days on simple. Not 8 days across the board - that’s Progressive’s auto-focused book. But cutting mid-market cycle time in half is realistic when the rules layer sits between FNOL intake and the adjuster workbench.

I recommend CCOs stop measuring cycle time as a single number and start measuring it as three separate lines: simple STP claims (target 24-48 hours), complex adjuster-touched claims (target 5-10 days), and litigated / disputed claims (target managed through legal ops, not cycle time metric).

NPS lift from cycle time compression: +15-25 points on Bain’s research baseline, which drops 2-4% of book churn on typical mid-market retention math.

Pain: LAE is creeping up and CEO wants it under 10%

LAE typically runs 10-12% of premium at mid-market carriers. Progressive runs closer to 7%. The gap is where CFO conversations get uncomfortable.

I’ll be direct: no single technology move drops LAE by 3 points. What moves LAE is compounding operational efficiency across the claim lifecycle. Automated triage eliminates the first handoff. Rules-driven coverage check eliminates the second. Rules-driven reserve calculation eliminates a third. Rules-driven settlement approval eliminates a fourth. Each handoff removed cuts LAE by 0.2-0.5 percentage points. Five handoffs removed cuts LAE by 1-2 points. Which is exactly the realistic mid-market target: 10-12% → 8-10%.

On a $2B GWP book, that’s $20-40M annually in LAE reduction. That’s the CFO number. That’s what closes budgets.

The catch: LAE reduction only sticks if the rules layer actually removes work, not adds it. I’ve seen carriers deploy BRMS platforms as parallel systems where adjusters still touch every claim to “verify the rule result.” That defeats the purpose. The right pattern: rules approve or route, adjusters exception-handle. If STP approval rules require adjuster verification, you haven’t automated - you’ve added a review step.

I’ve worked with a $1.8B GWP carrier that stripped adjuster verification from simple auto glass STP path and dropped LAE 0.8 points in six months on that LoB alone. That’s what direct handoff removal looks like.

Pain: Claims STP rate is stuck at 30% and senior adjusters are quitting

This is the pain that surprises CCOs most. Adjuster turnover at mid-market carriers runs 22% annually in the segments I work in. Exit interviews consistently identify the same driver: 70% of an adjuster’s day is data entry and routine eligibility checks on simple claims. The senior adjusters - the ones with 10+ years of coverage expertise - quit fastest because they’re doing entry-level work.

Claims automation isn’t about replacing adjusters. It’s about protecting the senior ones from the work that’s driving them out. When rules handle the 60-75% of simple claims that don’t need coverage judgment, the senior adjuster’s day shifts to complex coverage disputes, litigation strategy, SIU coordination on suspicious claims, and customer relationship work on high-severity cases. The job becomes more interesting, not less needed.

I’ve seen turnover drop from 22% to 12-15% at mid-market carriers within 18 months of hitting stage 3 automation. That’s a talent retention math that CCOs don’t put on the ROI deck often enough. If you’re paying $85-110K annually per senior adjuster and losing 22% of them, recruit-and-train cost alone runs $30-40K per replacement. On a 300-adjuster claims org, that’s $2M+ in avoided churn cost annually - before LAE reduction and NPS impact enter the conversation.

I’ll be direct on the framing that kills these programs internally: the CCO who tells the claims org “we’re replacing adjusters with AI” loses the team within six months. The CCO who tells them “we’re removing the work that’s driving our best people out” keeps them. Same technology. Very different program outcomes.

Pain: CAT response overwhelmed the last hurricane and reinsurer noticed

Every mid-market carrier with meaningful hurricane, wildfire, or convective storm exposure faces this pain. Normal week: 500-1 500 claims. Hurricane week: 50 000 claims in 48 hours. The system that handles the normal load doesn’t scale 100× on demand.

The CAT scalability requirement isn’t marketing math. It’s what your reinsurer notes in the renewal cycle when claims processing bottlenecks become the loss driver. I’ve seen carriers get flagged in reinsurer submissions for “business interruption from claims system bottleneck” - which raises reinsurance cost across the whole book, not just CAT-affected regions.

The architectural constraint: sustained request throughput under peak load. Higson runs at 9 000 requests per second per node, which translates to 32M requests per hour of sustained throughput. Hurricane surge of 50 000 claims in 48 hours works out to ~1 000 requests per second including redundancy - well within a single-node deployment. Legacy claims platforms that struggle at 5 000-8 000 requests per hour collapse under this load.

The CAT playbook that works in practice: rules-driven triage on the FNOL surge (auto-route simple property to STP, escalate complex to adjuster queue, flag potential fraud for post-CAT review), rules-driven reserve pre-calculation (reserves set within 24 hours of FNOL on rules-approved severity), rules-driven communication automation (customers get status updates without adjuster touch), rules-driven adjuster load balancing (route to adjusters by geography + specialty + current workload).

BNP Paribas Cardif isn’t a CAT-heavy carrier, so I anchor the CAT conversation on anonymized deployments. What I can say from the public reference: same BRMS layer, same architecture, same performance characteristics. The centralized rules platform is what makes the CAT surge tractable - regardless of which weather event drives it.

Pain: Settlement accuracy is 82% and rework is costing $8M annually

Internal audits at mid-market carriers commonly find 15-20% of closed claims have accuracy errors requiring reopening: rule misapplication, missed coverage, math errors on deductibles or depreciation, subrogation opportunities missed, salvage under-recovered. On a $2B GWP book, this typically costs $8-15M annually in direct rework plus 3-5 NPS points from re-contact of customers whose claims were closed incorrectly.

The rework rate is a rules problem. Adjusters applying policy rules manually across 50-state variation, multiple LoBs, and endorsement complexity make errors. The literature on human error in complex rule application is unambiguous - error rates run 8-15% on high-complexity tasks. Rules engines applying the same policy rules deterministically don’t make those errors.

I’ve watched carriers cut rework rate from 18% to 5% within 12 months of moving coverage check and settlement calculation into a BRMS layer. On the same $2B GWP book, that’s $5-10M annually in avoided rework, plus the NPS lift from customers whose claims close correctly the first time.

The audit trail matters as much as the error rate. When a rule computes a settlement and stores rule version + policy version + input data + output amount in the audit log, you can answer “why did we settle this claim this way?” in one query. Adjuster judgment + email + spreadsheet notes doesn’t answer that question in a defensible way when Compliance walks in.

Pain: Litigation rate is growing and each litigated claim costs 3-4×

Auto BI litigation rates trended from 12% to 18% at mid-market carriers over the last five years across the segments I work in. Each attorney-represented claim costs 3-4× the non-litigated equivalent - direct attorney fees, extended cycle time, higher settlement amounts, litigation reserve impact.

The pattern I see: litigation entry correlates with cycle time and communication gaps. Claimants who wait 45 days for a coverage decision escalate to attorneys. Claimants who get transparent status communication and a fair settlement offer within 10 days rarely do. The correlation isn’t perfect - some claimants attorney-shop regardless - but the driver of preventable litigation entry is process failure, not fraud or unreasonable demands.

Rules-driven fast fair settlement reduces litigation entry. Not through settlement generosity - through cycle time compression and communication reliability. When a rule computes a fair settlement offer within 72 hours of complete FNOL and the automated communication flow keeps the claimant informed at every step, the attorney call rate drops.

CCOs should measure preventable litigation as a separate line from total litigation. Preventable litigation is claims where cycle time exceeded 30 days or communication frequency dropped below weekly touch. That’s the metric that rules-driven claims process actually moves. Total litigation includes fraud-related, coverage-disputed, and severity-disputed cases that need legal ops regardless of process.

Pain: NAIC market conduct exam flagged inconsistent settlement practices

The NAIC Unfair Claims Settlement Practices Act (Model #900) is the regulatory backbone here. Core requirement: the same claim, presented twice with the same facts, produces the same settlement decision. Adjuster discretion creates inconsistency. Rules create consistency.

I’ve watched a $1.4B GWP carrier receive a $300K NAIC market conduct exam fine from a state examiner for inconsistent auto BI settlement amounts across adjusters. The remediation order required documented decisioning consistency. The fix: externalize settlement calculation rules into a BRMS, capture per-claim audit trail showing the exact rule version applied. NAIC closed the exam in three weeks once the audit trail was live.

The fine was the visible cost. The invisible cost was the C-suite time absorbed by the examination - CCO, General Counsel, Chief Compliance Officer, and CFO all in weekly examination-response meetings for three months. That time cost multiples of the $300K.

State-by-state variation makes this pain particularly acute for multi-state carriers. Florida hurricane deductibles, Texas wind/hail markers, California earthquake endorsements, no-fault PIP in 12 states - each state’s coverage rules layer into settlement calculation. Adjuster judgment across 50 states can’t stay consistent by memory. Decision table inheritance in a BRMS (base rule + state overrides) handles it.

The audit trail is what makes the architecture defensible when the examiner asks “why did you settle this claim this way?” - rule version, rule author, deployment date, input data, output decision. Adjuster judgment plus emails doesn’t survive that question. Rule audit does.

Pain: Claims AI vendor promised 35% STP improvement, delivered 8%

This is the pain that keeps CFOs skeptical of the next claims modernization budget request. Mid-market carriers routinely spend $2-5M on claims AI vendors over 18-24 month engagements with disappointing STP outcomes. The pattern:

The vendor promised outcomes achievable only on greenfield deployments with clean data, uniform LoB, and unlimited change management runway. The carrier’s actual reality - legacy data quality, multi-LoB complexity, adjuster change resistance, PAS integration friction - cuts realized outcomes to a fraction of the pitch. Vendor blames “change management.” Carrier blames “vendor overpromise.” Both are partially right.

I’ll be honest about the pattern: claims AI without a stable rules foundation delivers single-digit STP improvement. Without deterministic rules underneath, the AI scores can’t be acted on reliably - the downstream rules to route, settle, and audit don’t exist yet. The AI produces a triage prediction, but there’s no rule to convert prediction into action. So the adjuster still touches every claim to “verify the AI recommendation.” Which is exactly the workflow you were trying to eliminate.

The pattern that works: rules foundation first (deterministic coverage, deterministic settlement calculation, deterministic routing), AI layered on top for prediction where deterministic rules can’t cover - triage prediction on ambiguous FNOL, damage severity prediction from photos, fraud pattern detection on complex claims. AI enhances the rules; rules aren’t replaced by AI. Same architectural pattern that wins in fraud detection, applied to claims.

The realistic budget math: $200K-$3M total program for mid-market BRMS deployment over 6-12 months, payback 12-18 months post-go-live. Anyone promising 6-month payback is under-scoping the rule migration effort. Anyone promising “AI-driven 90% STP” is under-scoping what STP actually means at mid-market complexity.

The BNP Paribas Cardif reference - a claims consolidation that hit all eight

The BNP Paribas Cardif deployment is the public reference I anchor on because it illustrates the eight-pain pattern at scale.

BNP Paribas Cardif had separate claims processing systems for banking products (mortgage life insurance, credit protection) and traditional insurance products. The CCO’s question was consolidation-driven: “Can we run both books on one BRMS layer?” The 8-month centralization program delivered:

Unified claims cycle time reduction across both books (Pain - NPS lift)

LAE reduction from consolidated operations (Pain - CFO metric)

Adjuster capacity reallocation across products (Pain - talent retention)

Single audit trail across both regulators - banking and insurance (Pain - compliance)

Consistent settlement calculation across products (Pain - rework reduction)

The unexpected outcome the CCO surfaced post-go-live: cross-vertical fraud detection. The SIU team caught a fraud ring claiming on both banking products and insurance products separately for 18 months. Siloed systems missed it. Unified rules layer caught it. The SIU lead told me: “We were always going to centralize for efficiency. We didn’t expect fraud detection would be the bigger win.”

That’s the pattern. Centralize on the BRMS layer for the eight-pain CCO reasons. The compounding wins - cross-vertical fraud detection, unified analytics, rule reuse, single audit trail across regulators - emerge as ROI multipliers after go-live. The COO who built the program got the CFO business case on the deck. The CCO got everything on this article’s list plus a $4M+ fraud loss prevention win as bonus.

What you actually buy - the BRMS layer underneath claims modernization

Quick technology stack orientation for CCOs who don’t want to become architects.

Policy Administration System (PAS). Guidewire ClaimCenter, Duck Creek Claims, Insurity, Sapiens. These are the enterprise claims workbenches - adjuster UI, case management, document storage, workflow. PAS is your operational center. Higson doesn’t replace these - we integrate as the BRMS decision layer underneath.

Business Rules Management System (BRMS). Where externalized claims rules live - the claims rules engine handles coverage rules, reserve calculation, settlement rules, and triage rules. Higson sits here. So do alternatives like InRule, IBM ODM, FICO Blaze. The BRMS is where 60-75% of the operational levers in this article actually execute.

AI/ML layer. ONNX-deployed models for triage prediction, damage estimation, document extraction, fraud scoring. Runs inside the BRMS (Higson pattern) or as a separate microservice.

Document AI. OCR + NLP layer for claim documents - police reports, medical records, damage estimates. Specialist vendors (Hyperscience, Rossum, AWS Textract) or cloud platform builds.

Workflow orchestration. Camunda, AWS Step Functions, or vendor-embedded orchestration inside the PAS.

Honest Higson positioning: Higson is a BRMS for mid-market P&C carriers running enterprise PAS suites who’ve identified the BRMS layer as the bottleneck. We don’t replace Guidewire ClaimCenter or Duck Creek Claims. We integrate as the specialized rules-execution and governance layer. For carriers running lighter-weight claims workbenches or custom-built portals, Higson can be the primary rules layer underneath. We’re not the right fit for carriers looking for a full PAS replacement - that’s a different procurement decision.

For carriers with meaningful multi-line exposure (banking + insurance products), the same Higson BRMS handles both - the BNP Paribas Cardif cross-vertical pattern.

FAQ

What is a realistic claims STP rate for mid-market carriers?

For mid-market P&C carriers ($500M-$5B GWP), realistic claims STP rates are 60-75% on simple claims (auto glass, minor property, simple PIP) and 30-50% blended across all claim complexity. Specialty lines like auto glass can reach 90%+ STP. Anyone promising 100% STP across all claim types is selling marketing fiction - complex bodily injury, large commercial property, and litigated claims require adjuster judgment regardless of platform.

How much does claims modernization reduce LAE?

Mid-market P&C carriers typically see LAE reduction of 1-2 percentage points, from 10-12% down to 8-10% of premium, through compounding handoff removal across the claim lifecycle. On a $2B GWP book, that translates to $20-40M annually in LAE reduction. The reduction only sticks if the rules layer actually removes adjuster work rather than adding verification steps - the most common implementation failure.

How long does it take to implement claims modernization?

Mid-market BRMS-based claims modernization typically takes 6-12 months for a single line of business and 12-24 months for full multi-line deployment. Total program cost runs $200K-$3M depending on scope. Payback periods run 12-18 months post-go-live. Enterprise PAS replacements are separate 24-36 month programs and shouldn’t be confused with claims modernization.

How does claims automation affect adjuster jobs?

Claims automation handles 60-75% of simple claims (routine coverage checks, deterministic settlement calculations) without adjuster touch, freeing adjuster capacity for complex coverage disputes, litigation strategy, SIU coordination, and high-severity claims work. Adjuster turnover typically drops from 22% legacy to 12-15% with modern automation - talent retention is a strategic outcome, not a side effect. The framing matters: “we’re removing routine work” keeps the team; “we’re replacing adjusters with AI” loses them.

What does NAIC Unfair Claims Settlement Practices Act require?

The NAIC Model Act #900 requires consistent claims handling across states - the same claim with the same facts should produce the same settlement decision. It mandates timely communication with claimants, fair settlement offers, and documented decisioning. Rules engines support compliance by producing audit trails per claim showing the exact rule version and inputs that produced each decision - the audit format NAIC market conduct examiners review.

How does rules-based claims processing reduce litigation rates?

Rules-driven fast fair settlement reduces preventable litigation entry - claimants who receive coverage decisions within 72 hours and settlement offers within 10 days rarely escalate to attorneys. Cycle time compression and communication reliability are the drivers. Rules-based claims processing doesn’t reduce litigation on fraud-related, coverage-disputed, or severity-disputed cases - those need legal ops regardless of process. CCOs should measure preventable litigation as a separate line from total litigation.

How do rules engines support claims automation?

A claims rules engine externalizes claims business logic - eligibility rules, coverage triggers, fraud indicators, reserve calculations, settlement formulas - into a dedicated decision layer outside the policy admin system. Claims operations teams author rules in no-code interfaces; rules execute in milliseconds; every decision produces an audit trail. This combination enables consistent decisioning across adjusters, NAIC market conduct compliance, and rapid rule updates without IT cycles.

Does claims modernization need AI, or are rules enough?

Both. Rules handle deterministic coverage, deterministic settlement calculation, and deterministic routing - 60-75% of simple claims. AI enhances rules for prediction where deterministic logic doesn’t cover: triage prediction on ambiguous FNOL, damage severity from photos, fraud pattern detection on complex claims. Claims AI without a rules foundation delivers single-digit STP improvement because there’s no rule layer to convert AI predictions into action. Rules-first, AI-on-top is the architecture that produces meaningful STP.

Can the same rules engine handle claims and fraud detection?

Yes - for multi-line carriers, a shared BRMS hosts both claims processing rules and fraud detection rules. The BNP Paribas Cardif deployment is the public reference: the same rules engine runs claims rules and cross-vertical fraud detection rules across banking and insurance products. Inline fraud check inside the STP claims path requires sub-100ms latency, which BRMS platforms with sub-millisecond rule execution support.

Talk to Higson

If you’re a CCO facing three or more of the eight pains this article walks through - LAE creep, cycle time stuck at 25-30 days, senior adjusters quitting, CAT response fragility, NAIC market conduct findings, litigation growth, settlement rework at 15-20%, or claims AI vendor disappointment - that’s the operational reality we solve for weekly at mid-market carriers.

Schedule a 45-minute claims architecture review - walk through your carrier’s LAE, cycle time, and STP baseline against the mid-market benchmarks in this article; identify which of the eight pains your current stack addresses and which it leaves open; discuss the BNP Paribas Cardif cross-vertical pattern if you run multi-line products. Book a slot.

Self-serve a PoC - Higson is available on AWS Marketplace at $0.63/hour for proof-of-concept deployments. Start here.