Changes are coming for SNAP. Here’s how states can seize the moment.

In 2024, nearly 42 million Americans benefited directly from the U.S. Government’s Supplemental Nutrition Assistance Program (SNAP), according to the USDA. That means around 12% of the U.S. population depends at least in part on the $100 billion program for their food security. It’s safe to say that any changes to SNAP will have profound ripple effects across American society.  

Changes are coming. With the passage of House Reconciliation 1—the federal legislation also known as the One Big Beautiful Bill Act—individual states will have to shoulder 75% of SNAP’s administrative costs, up from the current level of 50%, starting in 2027.  

What’s more, in fiscal year 2028, states with a payment error rate of 6% or higher will be required to share the cost of SNAP benefits as well, with the amount determined on a graduated scale. Based on 2024 error rates, this new rule could result in an additional $11.6 billion in costs for states, according to Guidehouse analysis. 

Transformation needs to start now 

To navigate these changes successfully, states need to act quickly. Once the new cost sharing structures kick in, it will be that much harder for agencies to find money to pilot solutions for reducing SNAP errors and—just as critically—to address their root causes. 

Those causes are hardly a secret to SNAP eligibility workers, policy specialists, trainers, directors, and aligned health and human services (HHS) professionals, who increasingly report feeling overwhelmed and under-resourced.  

They find themselves having to carry out an emotionally demanding mission—helping put food on the table for families experiencing joblessness, disability, and other kinds of hardship—using outdated systems and manual review processes. The volume of cases and applications frequently exceeds capacity and strains quality controls, leading to staff burnout and increased turnover. 

Many factors contribute to the SNAP payment error rate. Among the most common ones are misreporting and inaccuracies surrounding wages and salaries, household composition, and household expenses. Persistent strains on SNAP staff and systems compound the problem, creating a perfect storm that impacts the accurate and efficient delivery of SNAP benefits to families in need.  

Errors, in turn, further erode morale across state agency staff, clients, and the public—and now pose a broader threat in the form of federally imposed financial penalties. SNAP staff urgently want better tools, better tech, and better collaboration.  

In that respect, the new federal rules represent an inflection point. By moving decisively to implement AI-enabled solutions and human-centered organizational change, HHS leaders can seize a generational opportunity to help SNAP staff fulfil their mission, improving the food security of millions of Americans. 

Reducing errors starts with technology 

Artificial intelligence (AI) and machine learning (ML) have the potential to dramatically improve error detection and prevention by identifying anomalies in SNAP cases before they reach an eligibility determination, or even quality control sampling, while also enabling near-real-time corrections.  

These capabilities are game changers—not just for states’ overworked eligibility workers, supervisors, managers, policy teams, audit teams, fair-hearing teams, and executives, but for the thousands of families whose nutritional health depends on the accurate and timely delivery of SNAP benefits. 

AI and ML solutions can identify case authorizations for error probability using both deterministic and probabilistic models. Those models, in turn, can learn and improve over time using an integrated feedback loop, allowing error sources to be identified at the case level so that teams can discern patterns and trends and take action to correct them.  

What’s more, AI-enabled error probability detection is explainable, transparent, and auditable, empowering workers to build trust and strengthen accountability with stakeholders both within and outside the agency.  

Piloting a solution that can integrate with a state agency’s eligibility systems should follow a phased approach: 

• Phase 1: Setup. Implementation teams formalize the pilot objectives and desired outcomes and analyze existing SNAP processes, data definitions, and data availability, as well as the technical environment in which analytics and model training and deployment will be performed.  
• Phase 2: Evaluation. Teams review, evaluate, and profile the provided data to determine the appropriate modeling and analytic approaches.  
• Phase 3: Configuration and optimization. The modeling and analytic approaches are configured to the designated pilot population, and potential model adjustments are made based on target performance and scoring thresholds.  
• Phase 4: Validation and testing. Teams finalize the model error detection triggers and select appropriate risk score thresholds, as well as confirm the plan for deployment.  
• Phase 5: Deployment and monitoring. The pilot launch should be accompanied by rapid incident detection and correction, with results compared against the established success criteria. Learnings will inform the expansion beyond the pilot population.

A successful, phased implementation of AI-enabled error detection and prevention is only part of the equation, however. New technology won’t move the needle for overburdened workers unless the solutions are paired with systemic, people-centric change. 

 

Taking a holistic view  

Meaningful change happens when leaders adopt a holistic—and humanistic—approach to reducing errors and find multiple ways to boost operational efficiency and collaboration across agencies and departments. That starts with three critical actions: 

1. Get buy-in for the new tech. AI and ML solutions won’t start delivering error-reduction and efficiency dividends unless workers know how to use the new tools and are incentivized to do so. Engage case managers and eligibility workers before deployment. Understand their pain points and day-to-day requirements. When people have a hand designing and configuring a solution, they’re more motivated to use it.  
   
   
2. Change processes and ways of working. AI-enabled tools can catch errors, identify patterns, and even correct mistakes—all capabilities that can ease the burden on reviewers, case managers, and other workers.  
   
   But those tools alone can’t fully address the root causes of SNAP errors, which can stem from siloed operations and inefficient workflows. Leadership teams need to apply business process re-engineering best practices to locate bottlenecks, identify redundancies, close skills gaps, and improve collaboration between departments and functions.  
   
   
3. Encourage communication and foster respect. You can’t effectively address workers’ pain points if you don’t have a clear picture of what they are. Better communication is key. Open-door policies, one-on-one meetings, listening sessions—these are a start. Establish two-way feedback mechanisms that allow workers to assess leadership effectiveness and close perception gaps. Being heard fosters a sense of mutual respect.  

These broad areas of action can manifest in myriad specific ways, from shared-solution databases that leverage the on-the-ground know-how of case managers to new centers of excellence focused on maximizing the potential of AI solutions. Always, the solutions should be informed by the experience of those closest to the work. 

Looking beyond the near-term benefits 

The urgency of investing in error-reduction technology is clear, as are the immediate benefits—namely, significantly easing the cost-share burden imposed by the enactment of H.R. 1. But to fully seize the moment, HHS agencies should anticipate, and work toward, a broader range of outcomes, including: 

• A newly empowered workforce: Significant reduction in staff time spent auditing cases can free case managers and other HHS professionals to focus on higher-order work such as training, stakeholder engagement, and long-term culture-building within the organization. 
• Improved reporting: Clear, explainable data and insights on error rates and improved efficiency can build trust with federal partners. 
• Lower administrative costs: In addition to avoiding the burden of cost-sharing, agencies stand to reap further dividends in the form of reduced overhead and related costs. 
• Fewer overpayments and fair hearing appeals: Accurate processing reduces disputes, easing stress on both administrators and beneficiaries. 
• Better outcomes for citizens: Faster and more accurate case processing improves overall service delivery. 

Other potential follow-on effects of implementing error-reduction solutions may become apparent only over time. What’s clear is that the moment to act is now, before the new federal rules take effect and potentially constrict resources further.  

By embracing a program of technological and human-centric change, states can emerge from this major policy shift with newfound agility and resilience.