A state Department of Health and Human Services managing Medicaid, SNAP, CHIP, and TANF benefits for 1.4 million residents was failing the people it existed to serve. Benefits processing averaged 47 days—far exceeding federal timeliness standards and triggering $18M in annual federal penalty assessments. The system ran on a COBOL-based mainframe purchased in 1992, operated by a shrinking pool of aging specialists, and accessed by caseworkers through green-screen terminals that required 6 weeks of dedicated training just to navigate. Residents had no way to check their application status online. The department was trapped between a system it couldn't afford to keep and a transformation it couldn't afford to get wrong.
Government & Public Sector
State Benefits Administration Digital Transformation
Primary Outcome
Cut benefit processing time from 47 days to 12 days and saved $34M annually while serving 1.4 million residents faster
12 days
Processing Time
$34M
Annual Savings
94%
Digital Adoption
14 Mo
Implementation
Project Overview
The Challenge
1. Federal Timeliness Violations & Penalty Exposure
Federal regulations require Medicaid eligibility determinations within 45 days and SNAP determinations within 30 days. The department was averaging 47 days across all programs—consistently violating both thresholds and accumulating $18M in annual federal penalty assessments. Each penalty dollar represented a dollar unavailable for direct client services. CMS had issued a Corrective Action Plan requiring measurable improvement within 18 months or risking a federal takeover of program administration.
- $18M annual federal penalty for timeliness violations
- 47-day average processing time vs 30–45 day federal requirements
- CMS Corrective Action Plan with 18-month compliance deadline
2. 1992 COBOL Mainframe on Life Support
The eligibility system ran on a mainframe environment with fewer than 8 COBOL developers in the state capable of modifying it—3 of whom were within 3 years of retirement. The system's data model had 40,000+ tables accumulated over 30 years of patch-on-patch development, with no documentation for 60% of the business logic. Any modification risked cascading failures in programs serving vulnerable populations who depended on uninterrupted benefit continuity.
- 8 COBOL-capable developers remaining statewide, 3 near retirement
- 40,000+ database tables with 60% undocumented business logic
- $8.4M annual mainframe licensing and maintenance cost
3. No Digital Self-Service for Residents
Residents applying for benefits had two options: mail a paper application (average 12-day mail processing lag before work began) or visit a county DHHS office in person during business hours. There was no online application, no status check capability, and no digital document submission. The department's call center handled 35,000 monthly calls, the majority from residents asking 'where is my application?' A single status inquiry required a caseworker to look up the record and manually read it to the caller.
- Zero online services—paper mail or in-person only
- 35,000 monthly status inquiry calls consuming caseworker capacity
- Average 12-day processing delay before paper applications entered the system
4. Manual Eligibility Determination Process
Caseworkers manually reviewed every application against program eligibility rules—a process that required cross-referencing 8 external data sources (SSA, IRS, DMV, wage records, immigration status, asset databases) through separate green-screen terminals. A single SNAP application required an average of 4.2 hours of caseworker time. Errors in manual determination created incorrect denials and overpayments, driving further appeal workload.
The Solution
API-Wrapped Mainframe with Modern Front End
Rather than replacing the mainframe's business logic—a risk too great given its complexity and the dependency of 1.4 million residents on continuous benefit issuance—we wrapped it in a modern API layer using strangler fig architecture. The 40,000-table data model remained unchanged; all new functionality was built against well-defined API contracts that translated modern service calls into mainframe transactions. This preserved 30 years of accumulated business logic while enabling modern application development above it.
Mainframe API Wrapper
REST APIs exposing 180 mainframe functions without modifying core eligibility logic
Strangler Fig Migration
New modules built against APIs; mainframe decommission planned for Phase 2
Resident Self-Service Portal
Launched a mobile-first benefits portal enabling online application for all four programs, real-time application status tracking, secure document upload, renewal management, and change-of-circumstance reporting. Multilingual support covers the 8 most common languages in the state. Application design followed USDS digital service standards with usability testing conducted with actual benefit recipients including elderly residents and people with disabilities.
- Online application for Medicaid, SNAP, CHIP, and TANF
- Real-time status tracking eliminates status inquiry calls
- Document upload replacing mail-in requirements
- Multilingual support in 8 languages with auto-detect
- WCAG 2.1 AA accessibility compliance tested with assistive technology users
Automated Eligibility Rules Engine
Built a modern eligibility rules engine that executes determinations for 85% of standard cases automatically, integrating with SSA, IRS wage data, DMV, and state wage records via secure API connections. The rules engine completes a full eligibility determination—including all 8 data source checks—in under 4 minutes. Caseworkers receive a pre-populated determination with all external data pre-filled; they review and approve rather than research and calculate.
- 85% of standard cases determined automatically in under 4 minutes
- 8 external data sources queried simultaneously via API
- Caseworker role shifted from data entry to determination review
- Audit-complete determination records generated automatically
Results & Outcomes
12 days
Average Processing Time
Average benefit determination time fell from 47 days to 12 days—a 74% reduction that places the department within federal timeliness standards for all four programs for the first time in 11 years. Fully online applications processed in automated track average 3.8 days.
2.1%
Determination Error Rate
Manual eligibility determination errors fell from 18% to 2.1% as automated rules engine replaced manual data lookup and calculation. Incorrect denials fell 84%, reducing appeal workload and ensuring eligible residents received benefits they had a right to.
$34M
Annual Operational Savings
Elimination of $18M federal penalties, $8.4M mainframe cost reduction (Phase 1 partial), reduced call center volume ($4.2M labor savings), and productivity gains from automated eligibility determination generated $34M in annual measurable savings.
94%
Digital Adoption Rate
Within 90 days of launch, 94% of new benefit applications were submitted digitally. Community outreach, library kiosk deployment, and caseworker-assisted digital navigation drove adoption even among residents without home internet access.
58%
Call Center Volume Reduction
Monthly status inquiry calls fell from 35,000 to 14,700 as residents checked their own status online. Caseworkers previously dedicated to answering status calls were redeployed to complex case management—improving service quality for residents with complicated circumstances.
4.6/5
Resident Satisfaction Score
Post-application satisfaction surveys rated the new digital experience 4.6 out of 5. The most frequently cited improvements were 'knowing where my application is' and 'not having to take time off work to come in person'—outcomes that reflect the equity impact of accessible digital government.
Technologies Used
Application Platform
React.js (Frontend)Node.js (API Gateway)PostgreSQLAWS GovCloud
Mainframe Integration
IBM Z API ConnectMicroFocus COBOL APIsIBM MQ Message Broker
Eligibility & Data
SSA / IRS Data Hub APIsState Wage Record APIDrools Rules EngineCamunda BPM Workflow
Business Impact
1.4 Million Residents Served Faster
Every resident who applies for Medicaid, SNAP, CHIP, or TANF now receives a determination in 12 days on average—a reduction from 47 days that meaningfully changes lives. For a family facing food insecurity, 35 fewer days of waiting is not an operational metric—it is a measure of whether children go hungry while the government processes their paperwork.
$34M Redirected to Direct Services
The $18M in eliminated federal penalties alone, redirected to benefit program spending, funds an estimated 4,200 additional Medicaid beneficiaries annually. Total operational savings of $34M represent resources returned to the mission of the department rather than consumed by the overhead of a failing system.
Foundation for Ongoing Modernization
The API wrapper architecture created a sustainable modernization path that the department's own technology team can continue executing. Phase 2—full mainframe decommission and modern rules engine expansion—is underway using state staff trained during Phase 1, reducing ongoing consulting dependency. The COBOL knowledge transfer program documented 94% of previously undocumented business logic.
Quick Project Info
Industry
Government & Public Sector
Services
Legacy Modernization, Digital Services, Process Automation
Duration
14 months
Client Overview
About the Client
A state Department of Health and Human Services administering Medicaid, SNAP, CHIP, and TANF benefits to 1.4 million residents across 100 counties through a workforce of 2,800 state employees and 4,200 county-level caseworkers.
Initial Situation
1992 COBOL mainframe on life support, 47-day processing times violating federal standards, $18M annual regulatory penalties, zero digital self-service for residents, and 35,000 monthly status inquiry calls consuming caseworker capacity.
Start Your Transformation
Ready to achieve similar results? Let's discuss how we can help transform your business.
Schedule Consultation Explore Services