What SNAP Quality Control Actually Measures
SNAP's quality control system measures payment accuracy on two dimensions: whether eligibility was determined correctly and whether benefit amounts were calculated correctly. Both overpayments and underpayments count toward a state's payment error rate (PER). Under 7 CFR Part 275, state agencies draw a sample of their active SNAP caseload each year and review those cases for accuracy. USDA then reviews a subsample of those same cases to validate the state's findings.
Errors below a tolerance threshold are excluded from the rate calculation. For FY2026, that threshold is $58, adjusted each year based on changes in the Thrifty Food Plan under Section 4019 of the 2014 Farm Bill. For example, in a hypothetical case, if a reviewer finds that a household received $60 more than it was entitled to, that difference would enter the error rate; if the hypothetical discrepancy were $55 instead, it would not — though the state would still be required to correct it.
SNAP payment errors are largely unintentional, as USDA has stated explicitly. They arise from incorrect application of eligibility rules, stale income or household data, procedural failures in documentation, and rule complexity when a household's circumstances change between certifications. The QC system is designed to measure administrative accuracy, not fraud. That distinction matters when designing corrective action: a system that surfaces fraud signals addresses a different problem than a system that catches benefit calculation errors from outdated household income data.
What QC does not measure is also important for state IT leaders to understand. Improper denials — cases where an eligible household was wrongly rejected — do not appear in the payment error rate. Neither do timeliness failures. The metric is a partial view of system performance.
The Financial Stakes Under H.R. 1
Starting with FY2028 (October 1, 2027), states face cost-sharing requirements based on their payment error rate:
- Below 6 percent: no state share; federal government covers 100 percent of benefit costs
- 6.0–7.99 percent: state covers 5 percent of benefit costs
- 8.0–9.99 percent: state covers 10 percent of benefit costs
- 10.0 percent or higher: state covers 15 percent of benefit costs
States whose FY2025 rate was at or above 13.33 percent receive a one-year delay; their cost-share begins with FY2029, not FY2028. Under current law (H.R. 1 §10105), a state whose FY2026 rate is independently at or above 13.33 percent is delayed to FY2030, regardless of its FY2025 rate. Neither delay is an exemption — those states face the same cost-share tiers when they apply.
Based on FY2025 state-level data, nine states and the U.S. Virgin Islands fall below 6 percent and will avoid the cost-share entirely in FY2028. Six states face the 5 percent tier. Sixteen face 10 percent. Thirteen states and Guam face the 15 percent tier immediately, with six additional states and the District of Columbia in the delay category but at risk of significant costs in FY2029 or FY2030.
The fiscal exposure varies widely by state caseload. Illinois's FY2025 rate of 14.67 percent qualifies it for the one-year delay, so cost-sharing would begin with FY2029, not FY2028, under current law. The Civic Federation's analysis estimates potential exposure at approximately $550 million annually, based on 1.565 million enrollees as of May 2026 at an average monthly benefit of $195 per person — if Illinois's rate falls below 13.33 percent but remains at or above 10 percent when cost-sharing applies. The scale communicates why error rate management is now a budget priority, not a program integrity footnote.
For the FY2028 first-year cost-share, states may elect the lower of their FY2025 or FY2026 rate; FY2026 rates will be published by June 2027. A hypothetical state that entered FY2026 at 11 percent but improved its accuracy during that year may benefit from the election. Congress is also considering changes: the Senate Agriculture Committee advanced S. 5526 (Agricultural Act of 2026) 12–11 on September 16, 2026 and formally reported it on September 24, 2026. If enacted, that bill would delay the cost-share to FY2029, base it on the lower of the FY2026 or FY2027 rate, require a rate of 13.33 percent or higher in both FY2025 and FY2026 for the FY2030 delay, and raise the top tier to 20 percent from FY2031. As of this writing S. 5526 has not passed the Senate; the CRS analysis of H.R. 1 and S. 5526 provides the current comparison. State planning should account for both current law and the pending Senate proposal.
Why Eligibility Systems Produce High Error Rates
SNAP eligibility systems were built as transaction engines. Their job is to process applications and recertifications, apply eligibility rules and issue benefits at scale. Most state SNAP systems accomplish that task — they move cases through a defined workflow. The accuracy problem is a different problem.
Payment errors generally arise from several compounding sources:
Stale data at certification. When a household's income or composition changes between certifications, the system may continue to calculate benefits based on outdated information until the next scheduled review. If a household reports a job change but the report is delayed or not processed promptly, the error accumulates each month.
Policy rule complexity. SNAP eligibility rules involve dozens of interaction effects — income limits that change with household size, deductions that vary by circumstance, categorical eligibility rules that differ by state. When state policy changes, updating eligibility calculation logic in legacy systems is not always instantaneous or complete. A rule change that was implemented correctly in the primary eligibility module may interact unexpectedly with the deduction calculation or the categorically eligible pathway.
Documentation processing gaps. Errors often occur when verification documents are required but not fully integrated into the determination. A paper income statement that is scanned but not read by the system — or read by a caseworker who processed it differently than the rule requires — becomes an error in QC review.
Caseworker workload and training. The FNA Keys to Payment Accuracy guide (published by FNS, now FNA) identifies staffing, workload management and organizational culture as core drivers of payment accuracy alongside data systems. A system that is technically correct but requires a caseworker to manually override multiple fields — hypothetically, seven separate steps to apply a single rule — will generate errors. System design that surfaced those overrides for supervisory review before benefit issuance would catch them.
These are system design and workflow design problems. They are not the same problem as fraud, and they don't respond to fraud-detection tools.
What a Corrective Action Plan Requires — and What Most Don't Include
Under 7 CFR 275.16(b)(1), state agencies with payment error rates at or above 6 percent are required to develop corrective action plans (CAPs). 7 CFR 275.17(b) requires CAPs to identify causal factors and assess the effectiveness of corrective actions. The FNA Keys to Payment Accuracy guide (published by FNS, now FNA) recommends that root cause analysis organize payment-accuracy drivers into the guide's defined keys: staffing and workload management, organizational culture, data systems and technology, training, quality assurance, policy, and corrective action planning.
Most CAPs that fail to reduce error rates share a structural problem: they diagnose symptoms rather than root causes. A plan that identifies "incomplete documentation" as a cause and prescribes "additional training" addresses the visible problem without changing the system conditions that produce it. Workers already know that incomplete documentation is a problem. The question is whether the system makes it easy or hard to surface that documentation before a determination is made.
A plan that has a realistic chance of changing error rates includes:
- A root cause analysis at the case level — not just category counts but specific pathways from a rule type to error generation, using QC case data.
- A system diagnostic that maps where errors originate in the workflow: at application, at recertification, at mid-period change reporting, or at benefit calculation.
- Measurable intermediate indicators — not just the annual PER, which is published roughly nine months after the fiscal year closes (around June, per 7 CFR 275.23(c)), but workload metrics, override rates, documentation completion rates that can be tracked monthly.
- Explicit technology requirements if system changes are part of the plan, with a realistic timeline. Eligibility-system projects that seek federal matching funds or approval go through the FNA Advance Planning Document process (FNA Handbook 901). That process exists for a reason — it governs how states document requirements, secure federal approval and access federal participation in system development — but it adds lead time that must be factored into a plan that targets a specific fiscal year's rate.
The Technology Investment Decision
Technology changes alone rarely move error rates significantly unless the root cause analysis points to a specific system failure. But when the root cause is in the system — policy rule calculation errors, data integration failures, missing real-time verification — process changes can only partially compensate.
The most actionable technology investments for state agencies facing high error rates generally fall into two categories.
The first is pre-issuance validation: systems that check a determination for consistency before benefits are issued, rather than detecting the error months later during QC review. This might be as straightforward as a real-time cross-check against wage data sources or a flag when a caseworker's income entry differs significantly from a submitted document. The technical requirements are modest, but they require integration with existing eligibility systems, which is where implementation risk concentrates.
The second is case risk stratification: identifying which cases are most likely to contain errors and prioritizing supervisory review of those cases before issuance. This requires historical QC data mapped to case characteristics — not an AI model with unexplained scores, but an explicit rule set or a well-validated classifier trained on the state's own error history, with transparent decision logic a supervisor can audit.
Both approaches require the same foundational work: clean data, defined integration points with the core eligibility system, and a deployment process that includes validation against the state's specific rule set before production use. Neither works as an off-the-shelf replacement for that integration work.
The Decision State IT Leaders Need to Make in FY2027
FY2027 is the current SNAP measurement year, running October 1, 2026 through September 30, 2027. Under current law, FY2027 performance sets the FY2030 cost-share. Under S. 5526 — if enacted — it could set FY2029. Either way, this is the measurement year in which operational changes made now can still affect the reported rate.
Eleven months is not long enough to complete a full eligibility system modernization. It may be long enough to implement targeted workflow changes, a pre-issuance validation layer, or a revised case management protocol for the error types that appear most often in that state's QC data. But only if the work starts now, with a root cause analysis that uses actual QC case data rather than categorical assumptions.
State human services technology directors and CIOs should ask their teams three questions: What specific rule types and case pathways appear most often in our QC error findings? Which of those are system-driven versus workflow-driven? And what APD process timeline would we need to initiate today to have a system change in production before FY2027 closes?
A CAP is already required at or above 6 percent. The cost-share starts at 5 percent of benefits for rates between 6 and 8 percent. The case for investment is straightforward at the numbers the FY2025 data shows most states carrying.
Sources and further reading
- USDA FNA announcement of FY2025 SNAP payment error rates (June 24, 2026)
- SNAP Quality Control — USDA FNA
- SNAP Quality Control Error Tolerance Threshold — USDA FNA
- 7 CFR Part 275 — Performance Reporting System — eCFR
- SNAP Keys to Payment Accuracy — USDA FNA (Food and Nutrition Administration)
- H.R. 1 SNAP provisions FAQ — Feeding America Action
- H.R. 1 and S. 5526 — CRS analysis IN12723 — Congressional Research Service (current version, Oct 5, 2026)
- New Data Underscore SNAP Cost Shift's Harm to Low-Income Families and State Budgets — CBPP (FY2025 state-level rates)
- Illinois Fails to Lower SNAP Payment Error Rates Targeted by H.R. 1 — Civic Federation
- The SNAP State Cost-Shift Policy Leaves the Program's Existence to Chance — Brookings/Hamilton Project
Spartan X builds the kind of data-quality and system-integrity capability that sits directly upstream of the problems a corrective action plan is designed to fix. The work of tracing an eligibility determination error to its system origin — identifying where a rule was applied incorrectly, where a data source was stale, or where a workflow step introduced inconsistency — requires the same discipline as any AI or analytics program: defined data lineage, documented decision logic and a validation process that doesn't wait for the annual QC report to surface a problem. Spartan X's state AI advisory and data architecture work is grounded in that same discipline, and it applies directly to program accuracy as well as to new capability deployment.
