Public Sector · Health and Human Services

Stellar ACCORD

Payment Accuracy for SNAP and Medicaid

ACCORD helps state and county agencies get eligibility and benefit decisions right the first time. It checks data before benefits are issued, connects records that live in separate systems, and gives caseworkers the support they need to work faster with fewer errors. The goal is simple: the right benefit, to the right person, at the right time.

Request a Briefing What H.R. 1 Changes

SNAP error threshold

Medicaid error threshold

Staff

Decide every case

Reported income differs from wage data

SNAP · Held for caseworker review

Review

Same person under two program IDs

Medicaid · Records linked for confirmation

Linked

Shelter deduction entered twice

SNAP · Corrected before issuance

Fixed

Medicare ID missing at application

Medicaid · Verification requested

Pending

Renewal documents complete and consistent

Medicaid · Cleared for determination

Clear

Illustrative display. No real case data.

Why Now

Payment Accuracy Now Has a Price Tag

The 2025 federal budget reconciliation law, H.R. 1, ties a state's share of SNAP and Medicaid costs directly to its error rates. Agencies that once absorbed errors as an administrative problem now face budget consequences.

10.62%

National SNAP payment error rate, FY 2025

Of the 50 states and DC at or above the 6% SNAP threshold in FY 2025

6.12%

Medicaid improper payment rate, FY 2025, against a 3% limit

75%

State share of SNAP administrative costs from October 1, 2026, up from 50%

SNAP: cost sharing tied to error rates

Starting in fiscal year 2028, states with a payment error rate of 6% or higher pay 5%, 10%, or 15% of their SNAP benefit costs, depending on how high the rate is. States below 6% pay nothing. SNAP improper payments totaled about $10.1 billion in FY 2025.

Medicaid: tighter eligibility error rules

Beginning in fiscal year 2030, the federal government must recover its share of eligibility-related payments above a 3% error rate, and waivers for good-faith effort end. In FY 2025, eligibility issues drove about 72% of Medicaid's $37.4 billion in improper payments.

Read the full H.R. 1 briefing, with dates, thresholds, and sources →

The Real Problem

Most Errors Are Process Errors, Not Fraud

CMS notes that an improper payment rate is not a fraud rate. About 77% of Medicaid's FY 2025 improper payments traced to insufficient documentation, meaning the file did not prove the payment was proper, which CMS says is generally not indicative of fraud or abuse. SNAP errors follow a similar pattern. They come from the way eligibility work gets done.

Systems that don't talk

Eligibility, benefits, and health plan systems each keep their own records. Decisions get made on stale or incomplete data, and one person can end up with more than one ID across programs.

Manual work under pressure

Caseworkers key in data, classify documents, and route cases by hand, often short-staffed and with rules applied differently from one office to the next.

Checks that come too late

Errors are usually found after benefits are issued, when quality control samples the case. By then the money is out the door and recovery is slow and costly.

How ACCORD Works

Prevent Errors Before They Are Paid

ACCORD sits alongside the systems an agency already runs. It adds checks, connections, and guidance at the points where errors start, and every flag comes with the reason behind it.

01 · Quality Gate

Check Before Issuance

A validation step reviews each determination before benefits are authorized. It catches calculation errors, duplicate deductions, and income that doesn't match available data, while there is still time to fix them.

02 · Identity and Records

One Person, One Record

Records for the same person are linked across programs and systems, so duplicate IDs, conflicting determinations, and missing Medicare identifiers surface early instead of turning into rejected claims later.

03 · Documents and Verification

Read It Once, Route It Right

Incoming documents are classified, key fields are extracted, and cases are routed to the right queue with a confidence score. Low-confidence items go to a person.

04 · Caseworker Support

Help for the People Doing the Work

Cases are prioritized by risk and deadline, forms are pre-filled from verified data, and the same rules apply in every county office. Caseworkers spend less time on keying and more on judgment.

05 · Program Integrity

Catch What Slips Through

Ongoing checks flag benefits that continue after a death, stale eligibility data, and unusual patterns, so they can be reviewed and resolved before they grow.

06 · Error-Rate Monitoring

Know Where You Stand

Leadership sees error trends against federal thresholds as they develop, by program, office, and error type, instead of waiting for the next quality control cycle.

Coverage

Built Around Where Errors Actually Happen

ACCORD was designed from a detailed review of documented error patterns in SNAP and Medicaid operations. Its capabilities are grouped into modules that address nine operational areas.

Operational area	What goes wrong	How ACCORD helps
SNAP benefit calculation	Wrong benefit amounts, duplicate deductions, errors at renewal	Pre-issuance quality gate and income cross-checks
Caseworker workflow	Backlogs, manual entry errors, uneven processing across offices	Case prioritization, pre-filled fields, consistent rules
Identity and duplicate IDs	One person with multiple program IDs, rejected transactions	Cross-program identity resolution and reconciliation
Medicare identifiers	Missing or incorrect Medicare IDs that cause claim denials	Verification before submission
Eligibility pathway gaps	Conflicting or duplicate determinations across eligibility groups	Cross-system eligibility checks before enrollment
Document processing	Misclassified documents, extraction errors, incomplete verification	Document classification, extraction, and routing with confidence scores
Program integrity	Payments after death, stale data, unexplained anomalies	Ongoing data matching and anomaly review
Provider payments	Late, incorrect, or duplicate provider payments	Pre-payment rate validation and early-warning alerts
Claims reprocessing	Mass rate corrections and retroactive eligibility changes	Repeatable, auditable reprocessing and reconciliation

Ten Modules, Three Phases

Start Where the Risk Is Highest

Each module can be adopted on its own. Most agencies begin with the modules that move the SNAP error rate fastest, then add operational and claims capabilities.

Phase 1 · Stabilize

Bring the error rate down

Eligibility Data Quality Engine
Identity Resolution and Record Management
Medicare Verification
Caseworker Workflow Accelerator
Error Rate Monitoring Dashboard

Phase 2 · Strengthen

Improve daily operations

Intelligent Document Processing
Cross-System Eligibility Intelligence
Provider Payment Accuracy
Fraud, Waste, and Abuse Detection

Phase 3 · Sustain

Handle large corrections

Claims Reprocessing Pipeline for rate changes, audit repayments, and retroactive eligibility

Access and Accuracy

Accuracy Without Wrongful Denials

Payment accuracy measures are lopsided. SNAP's payment error rate counts overpayments and underpayments but not improper denials or terminations, which USDA measures separately through its Case and Procedural Error Rate (CAPER). A tool tuned only to the error rate could lower it by pushing borderline cases toward denial. Every payment accuracy deployment, including ACCORD, should include safeguards like these.

Both Directions

Same Review for Denials

Proposed denials, terminations, and benefit reductions get the same review as approvals. A flag never denies a case on its own; it routes the case to a worker with the reason.

Measure Both

Track Denials Too

Denial and termination rates, procedural denials, and underpayments are monitored alongside the payment error rate.

Data First

Data Before Paperwork

When reliable data confirms eligibility, households are not asked for more documents, which cuts procedural denials and documentation errors together.

Equity

Watch for Uneven Effects

Shifts in denial rates or processing times across offices and population groups are flagged for review, consistent with USDA guidance on automation.

Our Principle

Augment. Don't Replace.

Caseworkers are not the problem. They are often understaffed and under-tooled. ACCORD is built to amplify their judgment, not substitute for it.

Every flag explains which data drove it
Eligibility decisions stay with agency staff
Works alongside existing eligibility and claims systems, with no rip-and-replace
Builds on the review practices agencies already use, such as second-party and supervisory review
Uses the results of existing data matches, such as the National Accuracy Clearinghouse, PARIS, SAVE, Social Security data, and state wage records, rather than replacing them
Rules and thresholds are configuration, not custom code, so the platform adapts to each state's policies
Modules can be adopted one at a time, based on an agency's own error profile

Security

Built for Sensitive Records

Benefit and health data demand strict controls, and ACCORD is designed around them.

Follows the permitted-use rules of each data source, including HIPAA, IRS Publication 1075, and Social Security and DHS SAVE agreements
Encryption at rest and in transit
Role-based access with full audit logging
Supports state requests for USDA approval of automation, with audit logs and accuracy monitoring
Delivered by Stellar, an ISO/IEC 27001:2022 certified, CMMI Maturity Level 3 appraised company

Who It Serves

Built for the Agencies Carrying the Load

State SNAP Agencies

Lower payment error rates ahead of federal cost-sharing.

State Medicaid Agencies

Stronger eligibility documentation before PERM and eligibility audits.

County Social Services Offices

Faster, more consistent case processing as administrative costs shift.

Managed Care Plans

Fewer rejected transactions and cleaner reconciliation after corrections.

Getting Started

From Assessment to Measurable Results

Error Profile Assessment

Agency + Stellar

Review recent SNAP quality control, CAPER, and Medicaid PERM findings and workflows to find where errors start.

Priority Modules

Stellar

Configure the modules that address the largest sources of error first.

Measure and Adjust

Agency + Stellar

Track error and denial indicators against federal thresholds and tune rules as results come in.

Expand

Agency + Stellar

Add operational and claims modules once the error rate is moving in the right direction.

Stellar ACCORD

Payment Accuracy Now Has a Price Tag

SNAP: cost sharing tied to error rates

Medicaid: tighter eligibility error rules

Most Errors Are Process Errors, Not Fraud

Systems that don't talk

Manual work under pressure

Checks that come too late

Prevent Errors Before They Are Paid

Check Before Issuance

One Person, One Record

Read It Once, Route It Right

Help for the People Doing the Work

Catch What Slips Through

Know Where You Stand

Built Around Where Errors Actually Happen

Start Where the Risk Is Highest

Bring the error rate down

Improve daily operations

Handle large corrections

Accuracy Without Wrongful Denials

Same Review for Denials

Track Denials Too

Data Before Paperwork

Watch for Uneven Effects

Augment. Don't Replace.

Built for Sensitive Records

Built for the Agencies Carrying the Load

State SNAP Agencies

State Medicaid Agencies

County Social Services Offices

Managed Care Plans

From Assessment to Measurable Results

Error Profile Assessment

Priority Modules

Measure and Adjust

Expand

Get Ahead of the 6% Line