A fiduciary intelligence layer is an independent, ongoing system of claims data review, TPA performance monitoring, and documentation that sits above a self-funded health plan’s third-party administrator. It exists because ERISA makes the plan sponsor, not the TPA, legally responsible for ensuring that claims are paid correctly.
A mid-size manufacturer with 650 covered employees paid the same six-figure inpatient claim twice. Nobody caught it for eleven months. The TPA's quarterly report showed 98% payment accuracy the entire time. That gap between what a TPA reports and what a plan sponsor actually owes under ERISA is the reason "fiduciary intelligence" is becoming a distinct discipline rather than a compliance afterthought.
What a Fiduciary Intelligence Layer Actually Is
A fiduciary intelligence layer is an independent, ongoing system of claims review and TPA performance monitoring that sits above the plan's day-to-day administration. Most employers assume their TPA's internal quality checks satisfy their oversight duty. In reality, TPA self-audits typically sample 250 to 400 claims out of tens of thousands processed annually.
That sample size covers a fraction of a percent of total claims volume. The remaining claims move through the system without independent review. A benefits committee that only reads the TPA's summary report is reviewing the TPA's opinion of itself, not the plan's actual financial and legal exposure.
The distinction matters because ERISA does not treat "we trusted our vendor" as a defense. [internal link: what is a fiduciary intelligence layer explainer] The fiduciary duty to monitor service providers is ongoing, not a one-time contract signature.
Why the Oversight Gap Exists
The gap exists because claims administration was built for throughput, not scrutiny. TPAs are compensated to process claims fast and keep denial rates low enough to avoid participant complaints. Neither incentive rewards catching an overpayment after the fact.
Volume compounds the problem. A plan processing 80,000 claims a year cannot be meaningfully checked by a quarterly spot audit, no matter how skilled the reviewer. Automation helps but does not close the gap. The 2025 CAQH Index estimates roughly $21 billion in remaining administrative automation savings across the industry, with claims processing still one of the least automated functions for complex, high-dollar claims.
Plan sponsors also underestimate how contractual language works against them. Service level agreements typically guarantee 98% accuracy against the TPA's own error definitions, not an independent standard. [internal link: TPA performance guarantees guide] A guarantee measured by the party being guaranteed is not oversight. It is a marketing figure.
The Real Cost of Skipping Independent Oversight
Willis Towers Watson puts the industry-standard TPA error rate at 1% to 3% of total claims processed, and some independent audits find error rates running higher on complex claim types. On a plan spending $10 million a year on claims, even the low end of that range represents real dollars leaving the plan every month. One Baker Tilly audit example found a TPA's actual financial accuracy at 96.8% and payment accuracy at 96.1%, both below the 98% SLA the contract required.
The cost is not only financial. Under DOL and EBSA guidance, fiduciaries who fail to monitor claims administration expose themselves and the plan to personal liability, not just plan-level penalties. EBSA closed 878 civil investigations in FY2025, with 63% producing monetary recoveries or corrective action, totaling $714.4 million.
A useful comparison: nobody would let a company's outside payroll vendor self-certify that every paycheck was correct with no independent check. Health claims run through a similar external vendor relationship, at a much larger dollar volume, with far less routine verification.
What's Actually Happening Behind the Scenes
Claims Leakage Hides in Ordinary-Looking Payments
Most overpayments are not fraud. They are duplicate payments, incorrect coordination of benefits, out-of-network claims paid at in-network rates, and specialty pharmacy claims billed above the negotiated rate. Each one looks routine in isolation.
Coding and Billing Drift
Upcoding and unbundling shift costs upward gradually, one claim at a time, without triggering any single red flag. A TPA's automated adjudication system approves claims that pass basic logic checks even when the underlying billing does not match the service rendered.
Eligibility and Dependent Drift
Dependents who should have been removed from coverage, such as an ex-spouse or an adult child who aged out, continue generating claims for months or years. Nobody outside a dedicated eligibility audit typically catches this pattern.
Why Current Approaches Aren't Enough
The status quo relies on the TPA to grade its own work once a year. A fiduciary intelligence layer replaces that with continuous, independent review built specifically to satisfy the plan sponsor's fiduciary duty rather than the vendor's contract renewal.
How to Build a Fiduciary Intelligence Layer
Red Flags That Signal Your Plan Needs This Now
The ROI of Getting This Right
A full independent claims audit typically recovers 1% to 3% of annual claims spend, according to industry benchmarking. On a $15 million plan, that is $150,000 to $450,000 in a single review cycle, before counting the ongoing savings from corrected processes going forward.
The fiduciary protection matters as much as the dollars. Documented, continuous oversight is the evidence a plan sponsor needs if EBSA opens an inquiry or a participant lawsuit alleges a breach of fiduciary duty. Kaiser Family Foundation's 2025 survey found that 67% of covered workers are now enrolled in self-funded plans, meaning this exposure applies to a majority of the employer-sponsored market, not a niche segment.
Employers that treat oversight as a line item rather than a favor from their broker tend to catch problems earlier and pay auditors less over time, because the error patterns get fixed instead of repeating every quarter.
Conclusion and Next Steps
Self-funded employers carry the fiduciary weight of every claim paid on their plan, whether or not anyone reviewed it. A fiduciary intelligence layer turns that exposure into a managed, documented process instead of an open question. The employers who build this now, before EBSA or a participant lawsuit forces the issue, are the ones protecting both their claims dollars and their fiduciary standing.
Start with one claims-level data pull from your TPA and a conversation with your benefits committee about when the plan was last independently audited. That single step usually reveals how much oversight is actually happening today.
Frequently Asked Questions
What is a fiduciary intelligence layer?
An independent, ongoing system of claims review and TPA monitoring that documents plan sponsor compliance with ERISA fiduciary duties.
Who is legally responsible for claims accuracy in a self-funded plan?
The plan sponsor, under ERISA Section 404(a)(1)(B), even though the TPA processes the claims.
How often should a self-funded plan audit its TPA?
Continuous or quarterly review is best practice; annual sampling alone leaves most claims unreviewed.
What percentage of claims typically contain errors?
Industry benchmarks from Willis Towers Watson put TPA error rates at 1% to 3% of total claims processed.
Can a TPA's self-reported accuracy be trusted alone?
No. Self-reported figures use the TPA's own error definitions and sample only a small fraction of claims.
What does EBSA look for in claims oversight enforcement?
Documented, independent monitoring of TPA performance, not just a signed administrative services contract.
How much can an independent claims audit recover?
Typically 1% to 3% of annual claims spend, based on published audit benchmarks.
Is claims oversight only relevant for very large employers?
No. Sixty-seven percent of covered workers nationally are in self-funded plans, spanning a wide range of employer sizes.
