Claims intelligence that traces error back to the benefit setup, contract load, fee schedule or edit that produced it, because a defect corrected once stops generating claims indefinitely and a recovery only retrieves one.
Payment integrity as an industry is oriented toward finding and recovering money after it has left, usually from provider billing patterns, usually for a percentage. That work has value. It is also downstream of a larger and cheaper opportunity: the plan own configuration, which generates error quietly, continuously and at scale, and which nobody is paid a contingency fee to find.
Recovering an overpayment costs a fee and a provider relationship. Preventing it costs a configuration change.
A vendor compensated as a percentage of recoveries has no commercial reason to identify the root cause that is generating the overpayments. Finding and fixing it would end the revenue stream. That is not an accusation of bad faith, it is a description of the arrangement, and the consequence is that plans accumulate recurring error patterns that are profitably harvested rather than corrected.
Internally the picture is similar. Payment integrity is measured on dollars recovered, configuration is measured on implementation delivery, and nobody is measured on payment error prevented. The measure does not exist in most plans, so neither does the work.
Contingency arrangements reward the symptom
A partner paid on recoveries is paid more when the underlying defect persists.
One configuration defect generates claims indefinitely
A wrongly loaded fee schedule, a benefit configured against the wrong plan or a missing edit produces error on every matching claim until somebody finds it.
Denial analytics stops at the reason code
Reporting denials by code describes the symptom. The useful question is which configuration, provider education gap or process failure produced that code repeatedly.
The correct claim is invisible
Analytics sees pends, denials, adjustments and appeals. Claims that processed correctly generate no signal, so the denominator for any error rate is usually assumed rather than known.
Auto-adjudication rate can be improved by removing edits
Reported together with payment accuracy it is useful. Reported alone it is an incentive to pay faster and less correctly.
An unusual claim is not an incorrect claim
Outliers produce investigation candidates, never payment conclusions.
Ask your payment integrity partner for their root cause report.
Not the recovery report. The list of underlying configuration, contract and process defects their findings trace back to, and what they have recommended you fix.
Individual claim errors are noise. Patterns are findings. The method is to group error by everything they have in common, because a cluster sharing a plan, a contract, a provider type, a service or a date range is almost always pointing at a single upstream cause.
Step 1
Establish the population of interest
Pends, denials, adjustments, appeals, overpayments and rework.participation on the service date is not participation today.
Step 2
Cluster by shared attributes
Plan, benefit, contract, fee schedule, provider type, service, place of service, edit and effective date range.
Step 3
Localize before decomposing
Establish where the change exists across line of business, product, geography, provider, facility, service, place of service and member cohort.
Step 4
Trace each cluster to its origin
Classify utilization, provider behaviour, contract, configuration, eligibility, coding, authorization, data, process or expected variation.
Step 5
Quantify the recurring cost
What the defect has cost to date and what it will continue to cost monthly if uncorrected.
Step 6
Separate plan-caused from provider-caused
The responses are entirely different and conflating them produces provider abrasion over the plan own errors.
Step 7
Prioritize prospective correction
A fix stops the bleeding and a recovery retrieves one instance at a cost.
Step 8
Route each finding
Configuration, contracting, provider relations, claims operations or payment integrity.
Step 9
Design prepay controls
Analyse leakage with capacity context, distinguishing patient or referral choice from the absence of an in-network option.
Step 10
Verify the fix and close the finding
Confirm the error rate fell after the change and track every finding through detected, reviewed, assigned, investigated, resolved and measured.
Separate the errors you caused from the ones you did not.
A substantial share of what payment integrity treats as provider billing issues traces back to the plan own configuration. Establishing the origin before acting prevents avoidable provider abrasion.
Detection is well served by the market. What is thinly served is prevention, root cause and the operational analytics around the claims that stopped, because none of those generate a contingency fee.
Prevent
Configuration Defect Analysis
Error clustered and traced to benefit setup, contract load, fee schedule, edit logic or effective dating, with recurring monthly cost quantified.
Prepay Control Design
Where a pattern is predictable, the control moved before payment rather than after.
Contract and Fee Schedule Validation
Expected versus actual reimbursement tested systematically.
Change Impact Analysis
The claims effect of a configuration, benefit or contract change modelled before release and monitored after.
Detect
Payment Accuracy Analytics
Claims paid incorrectly by amount, benefit application, coordination, duplicate or eligibility, with origin established before pursuit.
Provider Billing Pattern Analysis
Outlier identification with plan-caused patterns excluded first.
Duplicate and Coordination Detection
Duplicate payment and coordination-of-benefits error.
Underpayment Detection
Claims paid below contracted terms, which most integrity programmes never look for.
Operate
Pend and Denial Root Cause
Volume traced to the configuration, provider behaviour or process step generating it.
Rework and Adjustment Analytics
The cost of claims processed more than once, by cause.
Appeal and Dispute Analytics
Appeal volume traced to its origin.
Claims Operations Performance
Cycle time, touch count, first-pass rate and cost per claim by type and cause, reported alongside accuracy.
What CaliberFocus does, and does not do?
We are not paid on recoveries and will not accept an engagement structured that way, because a partner compensated on findings has an interest in the defect persisting. We also identify underpayments alongside overpayments, which reduces the headline number and is what accuracy actually means. And we will tell you when a pattern your integrity programme is pursuing as provider behaviour originates in your own configuration.
Most of these findings are reported as claims problems. The third column is where each actually originates, and it is the difference between a finding routed to the team that can correct it and one routed to the team that will process it again next month.
| Finding | How It Presents | Where It Usually Originates |
|---|---|---|
| Repeated Pends on One Edit | Volume in a claims work queue | Edit logic or configuration. Staffing the queue treats the symptom permanently. |
| Incorrect Reimbursement Amount | Overpayment or underpayment | Contract load, fee schedule or effective dating. Almost always a plan-side defect. |
| Benefit Applied Incorrectly | Member liability wrong, appeals follow | Benefit configuration, frequently against the wrong plan or product variant. |
| Duplicate Payment | Recoverable overpayment | Duplicate logic and provider resubmission behaviour together. Usually preventable prepay. |
| Coordination of Benefits Error | Paid as primary when secondary | Other coverage data quality and refresh cadence rather than adjudication logic. |
| Denials with High Overturn Rate | Appeal volume and provider abrasion | The original determination logic. A high overturn rate is evidence the denial was wrong. |
| Provider Billing Outlier | Pattern flagged for review | Genuine provider behaviour, or a plan configuration issue presenting as one. Establish which first. |
| Repeated Rework on One Claim Type | Adjustment volume and cost per claim | A defect somewhere upstream that nobody has traced because rework is not measured. |
Underpayment Is the Finding Nobody Is Paid to Look For
Claims paid below contracted terms are as much a payment accuracy failure as overpayments. Looking for both directions is what distinguishes payment accuracy from recovery, and it is also one of the most effective ways to improve provider trust in the claims operation.
The analytical technique that matters here is clustering error by every attribute the claims share and looking for concentration, because a defect expresses itself as a group of claims with something in common rather than as an unusual individual claim.
| Dimension | What Concentration There Suggests | |||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Plan, Product or Benefit | A benefit configuration defect, frequently introduced at plan year setup. | |||||||||||||||||||||
| Contract or Provider Group | A contract load or fee schedule error affecting every claim under that agreement. | |||||||||||||||||||||
| Service or Procedure | An edit, a code-level configuration issue or a coverage policy applied incorrectly. | |||||||||||||||||||||
| Place of Service or Facility Type | A setting-specific rate or rule configured against the wrong criteria. | |||||||||||||||||||||
| Effective Date Range | A change introduced on a known date. The most diagnostic dimension and the least used. | |||||||||||||||||||||
| Submitting Provider | Provider billing behaviour, or a provider record defect on the plan side. | |||||||||||||||||||||
| Edit or Reason Code | The logic itself, which may be too broad, too narrow or wrongly sequenced. |
| Category | What We Measure | Why It Matters |
|---|---|---|
| Error Prevented | Recurring monthly error eliminated by configuration and contract correction | The measure that does not exist in most plans, and usually the largest number. |
| Payment Accuracy | Claims paid correctly, in both directions, against a known denominator | Recovery measures what went wrong. This measures whether the operation is working. |
| Root Cause Closure | Findings traced to origin and corrected at source, against findings only recovered | Distinguishes a programme that fixes from one that harvests. |
| Rework Cost | Claims processed more than once, and the operational cost of that | A substantial expense that appears in no standard report. |
| Prepay Shift | Share of intervention occurring before payment rather than after | Every claim moved prepay removes a recovery, a fee and a provider conversation. |
| Provider Experience | Disputed findings, overturn rate on recoveries, and appeal volume from denials | The cost that arrives later, in contracting. |
