Executive Summary
The COVID-19 pandemic created an unemployment crisis across our nation, and Ohio was no exception. Almost overnight, there were more unemployment insurance (UI) claims in the Ohio system than in the entire previous year. Complicating Ohio’s response to unprecedented levels of claims, was a two-decades old legacy system, an entirely new federal unemployment program for gig-workers, never-before-seen levels of fraud, and a workforce sized for pre-pandemic levels of claims. Ohio’s successful response required integrating new technologies, processes, and people to navigate this emerging landscape. From quickly standing-up an entirely new system to manage the federal Pandemic Unemployment Assistance (PUA) program, to incorporating a comprehensive suite of fraud detection tools, to better routing and customer service tools, to creating a more flexible workforce, all while moving to a remote-working model, Ohio found ways to better serve unemployed workers.
This submission details Ohio’s situation and challenges, as well as its response. It includes detailing how Ohio managed multiple new and existing computer systems and leveraged existing and new technologies to better utilize data to prioritize and streamline processing. This was particularly important in stemming the flow of fraud and allowing us to focus on legitimate claims.
Ohio’s Unemployment System
The Ohio Department of Job and Family Services (ODJFS) is the state’s UI services agency. Ohio’s UI system is aging and slated to be replaced by a modern, cloud-based product. Due to legacy system limitations, the state contracted for a Software as a Service (SaaS) PUA solution when the federal government created this new benefit. The result was that as the pandemic began to unfold, Ohio had two UI systems, two contact centers (UI and PUA) and scores of supporting applications and business process.
Early on ODJFS recognized that potential fraud was clogging its systems, significantly delaying processing of legitimate claims. However, its hybrid UI system couldn’t generate data insights needed to effectively manage the crisis. ODJFS turned to the Ohio Department of Administrative Services (DAS) for help. DAS’ Innovate Ohio Platform (IOP) Data & Analytics Team was asked to build a fraud dashboard solution that would mine UI and PUA data to identify potentially fraudulent claims. This effort enabled ODJFS to identify questionable claims and focus first on claims that were more likely to be legitimate.
IOP then built a leadership dashboard. By bringing together ODJFS’ disparate UI and PUA operational data, IOP gave leadership needed information to accurately understand and report conditions, identify issues, improve business processes, and measure results. The impact was immediate and ongoing. With reliable and near real-time data, ODJFS was able to continually make course corrections and business process improvements that helped ODJFS better manage and align contact center and processing staff to meet the needs of the moment and expedited end-to-end claims processing.
Using an Agile project management approach, ODJFS was able to quickly leverage IOP’s extensive data ingestion, curation and analytics toolset and their deep advanced analytics expertise. This included machine learning technology that was able to effectively predict fraudulent claims, identifying hundreds of millions of dollars in fraud. Likewise, the leadership dashboard data allowed ODJFS to substantially improve business processes such as expanding contact center capacity by 65% to better serve Ohio citizens.
IOP and ODJFS continue to collaborate on this project, using lessons learned to inform the development of the new UI system, providing law enforcement with critical data to capture fraudsters, and expanding the dashboard solution approach across ODJFS’ other important human services programs.
Idea
The COVID-19 pandemic placed unprecedented demands on Ohio’s Unemployment Insurance (UI) program. In March 2020, the unemployment rate in Ohio was 4.8%. A month later, the rate soared to a staggering 16.4%. By May, there was a dramatic spike in new unemployment claims with 94,000 new claims filed in just one week. New claims reached an historic weekly high of 274,000 in late March 2020. In contrast, Ohio received about 6,200 new claims the last week of March 2019, and just under 230,000 new claims for the entirety of 2019.
The Ohio Department of Job and Family Services (ODJFS) is responsible for UI and Pandemic Unemployment Assistance (PUA). ODJFS has an aging UI legacy system scaled to handle normal transaction loads. Associated business processes such as the contact center, fraud detection and claims adjudication were similarly scaled. There was no corresponding infrastructure to receive and process PUA claims, and the legacy system and business processes could not be expanded to do so. Thus, ODJFS contracted for a PUA specific Software as a Service (SaaS) solution and contact center and had to establish separate business processes for PUA.
Squaring off against the crushing demand of Summer 2020, ODJFS had two separate and distinct unemployment infrastructures, both overwhelmed by the massive influx of initial claims activity. Systems slowed, the contact centers experienced long wait times and the adjudication backlog grew substantially. As ODJFS began to process these new claims, they soon realized they were being inundated by fraud. Potential criminals were flooding its systems, boxing out real claimants who needed immediate help. Many contact center calls, for example, were from citizens who had fraudulent claims opened in their name. The hybrid solution was unable to provide reporting that would inform ODJFS mitigation efforts.
ODJFS turned to the Ohio Department of Administrative Services (DAS) and its InnovateOhio Platform (IOP) Data and Analytics team for help. ODJFS asked IOP to combine data from UI and PUA, as well as other relevant state data, to create a fraud dashboard that would give ODJFS detailed insights into the issue. With this information, ODJFS was able to quickly identify potentially fraudulent claims and prioritize its attention to first handling those claims most likely to be legitimate. IOP then built a leadership dashboard giving ODJFS (close to) real time operational data, automating manual reporting that was providing error prone lagging indicators. This effort gave ODJFS the information needed to attack critical business process issues, improve efficiency, and better respond to citizens.
To make this work, ODJFS and IOP stood-up a solution in 30 days that would normally take a year or more. In addition to speed, this effort was notable in many other ways. Our fraud detection and prevention work became a template for other states facing similar problems. In fact, ODJFS provided technical assistance and best practices to several peer agencies, helping them accelerate their fraud prevention efforts. Consistent with 2022 State CIO Priorities, we also unlocked information at an enterprise level to quickly make decisions and course corrections, using data to optimize performance and modernize legacy processes and systems. Finally, this project showed the transformational power of data to improve government services. As the pandemic stabilizes, ODJFS continues to use the IOP dashboards to build out a sustainable UI infrastructure and business processes. They have also leveraged lessons learned to expand this data driven management approach across their other business lines and the entire enterprise.
Implementation
Together, IOP and ODJFS, under emergency conditions, knocked down many of the barriers that had previously stalled adoption of enterprise data utilization at ODJFS, including matching data elements from different systems and harmonizing business processes across separate programs. But this could only happen through robust project management and a unique staffing approach. The Governor and Lt. Governor immediately responded to ODJFS’ needs by initiating an innovative Public Private Partnership (P3). Comprised of twelve Ohio-based private sector companies, the P3 collaborated with ODJFS, sharing industry best practices on fraud detection and prevention, business process improvement, contact center responsiveness, and claims and adjudication streamlining. The P3 industry partners included financial services, processing, technology and advisory services, and insurance companies.
The P3 and ODJFS’ Data and Analytics organization immediately engaged IOP to determine the data needs, sources, requirements, and reporting needed to inform and direct enhanced fraud detection and prevention. Application Programming Interfaces (APIs) were built to ingest the required data into the IOP environment, which is a fully secure platform within Ohio’s state data infrastructure. The data came from the legacy UI system, the SaaS PUA solution, two contact centers, and associated support systems. The largest challenge IOP faced was harmonizing these various data elements to allow accurate and reliable analysis. Each system has a specific way in which data is captured, often using slightly different elements for the same data. At first, this created a situation where each system told a different story, but IOP was able to synchronize the various data elements to provide consistent results.
With the data fully curated and refactored, it was available for analytics and reporting use. IOP employed standard Tableau tools to visualize data for the Fraud dashboard. We then used Alteryx, a code-free data science (predictive, AI, ML modeling) technology, that is fully integrated within the IOP technical architecture. This was the key tool in maturing the fraud dashboard to accurately predict potentially fraudulent claims by discovering that many claims originated from the same IP addresses, used the same email addresses for contact and the same bank accounts for benefit payments. To ensure the fraud dashboard was producing accurate results, ODJFS tested it against a Google analytics solution and found that it was, indeed, precise.
With potentially fraudulent claims excluded, IOP helped ODJFS focus on improving service delivery to real claimants by building the leadership dashboard. This product is a comprehensive operational scorecard providing detailed visibility into claim numbers, claims paid, appeals filed and pending, contact center activity, adjudication backlog, workload, and pending issues. P3 helped ODJFS use this information to make targeted business process improvements that significantly expedited the payment of benefits to citizens in need. For instance, ODJFS could now prioritize workloads and flex resources up and down to meet demand. The contact centers were staffed appropriately at high volume call times and the adjudication backlog was targeted and effectively worked down.
The dashboard project was sponsored by the ODJFS Director and driven at the enterprise level by a P3 company adept at large-scale program management. Inputs included a demand side comprised of ODJFS UI leadership, P3 experts and support contractors who provided data, requirements, and analytic requests to the supply side, comprised of IOP leadership and staff. The ODJFS Data and Analytics organization managed day-to-day project work by intaking demand requests through a ServiceNow ticketing system and collaborating with the IOP team to scope, prioritize and complete tasks. A ODJFS UI Subject Matter Expert team was on call to assist IOP with program specific consultation.
Using an Agile approach, the program management team focused on creating the fraud dashboard. The Agile methodology was then used to build the leadership dashboard and to manage the ongoing maintenance and enhancements of both dashboards.
At inception, the project had a one criterion for success—identify and prevent fraudulent claims so that real claims could be more efficiently processed. Once this was accomplished, the next measure of success was to provide the right information at the right time to inform critical decision-making and business process improvements. Now, almost two years later, the project views success as providing data to inform the continual improvement of UI business processes and the stand-up of a modern cloud-based UI system under development as well as providing critical information to law enforcement to capture criminals and stolen funds. Just as important, success is measured by how well ODJFS, with IOP support, is using what was learned to continue down the path of data-driven business optimization for the rest of the enterprise.
Impact
This project allowed millions of vulnerable Ohioans to receive benefits faster, and with increasingly fewer problems navigating the end-to-end claims process. For individuals and families living paycheck to paycheck, these improvements were essential. IOP’s work also allowed ODJFS to identify hundreds of millions of dollars in potential fraud, particularly in the PUA program.
Data supporting the impact is truly astounding. Upon instituting new rules that flagged potentially fraudulent elements, the traditional UI system saw an 87 percent reduction in new claims within 60 days and the PUA program experienced a 98 percent reduction in initial claims over the same time period. In February 2021 alone more than 170,000 PUA claims were flagged as potentially fraudulent. Clearly, fraudsters were clogging ODJFS’ systems with claims at the expense of real people. Now, IOP has created a fraud dashboard interface with law enforcement so that they can consume this fraudulent claims data to be used in catching and prosecuting these criminals.
The leadership dashboard performance data was equally as impressive. For instance, early on this dashboard data was showing a bleak contact center situation. There just weren’t enough call center agents nor was the infrastructure scalable enough to handle the massive influx of calls. During peak periods the contact centers (UI and PUA) were receiving up to 250,000 calls per day. The wait times were long, citizens had to make repeat calls to get the right information and very few were able to utilize self-service options. ODJFS used this knowledge to institute immediate and targeted improvements. It significantly increased contact center capacity and capability by adding more trained agents, including ODJFS staff that were flexed from other duties. The data allowed ODJFS to schedule these agents so that the bulk would be available at high volume times during the day, and it directed the expansion of Spanish language agents, who were previously understaffed. Additionally, the data showed ODJFS that it needed to greatly expand its call center infrastructure to accommodate more calls and adjust its Interactive Voice Response (IVR) system to promote enhanced self-service. Over a three-month period from January to April 2021, the contact center was able to handle 65 percent more calls, the number of repeat callers was reduced by half and there was a 5 percent increase in callers using self-service.
Perhaps the best evidence supporting the impact of this project came from the Unemployment Modernization and Improvement Council (UMIC), the Legislative body created by law to oversee ODJFS’ efforts. In their August 2021 report, they said:
“As part of the on-going improvements with reporting, leveraging the Innovate Ohio Platform [IOP] …Enhancements with reporting helped to better show greater understanding of claims, fraud data, backlogs, data visualization, and a better display of the history of claims during the pandemic for all facets of the claims. The P3 appears to have dramatically improved the reporting capabilities for both OJI [UI] and PUA, as well as assisting in addressing fraudulent claims. Additional metrics for contact center focus, quality metrics, quality review standards, vendor Service Licensing Agreements (SLA), and additional standards have greatly improved over the course of the pandemic. Accuracy in reports and standard review processes have helped to better articulate the issues and the resolutions facing Ohioans in the Unemployment System. Standard metrics reported to the committee helped to demonstrate on-going evaluation and enhancements and build transparency with the general public.”
While the pandemic unemployment crisis has subsided, ODJFS continues to work with IOP in refining the dashboards and maintaining them for current and future insights. Recently, ODJFS and IOP began collaborating to integrate the dashboards with ODJFS’ new UI system scheduled to come online in 2023. The lessons learned during the pandemic will, therefore, be carried forward in continually improving ODJFS’ UI services capabilities.
Born out of necessity, this project was well worth the investment as it streamlined and expedited billions of dollars in legitimate unemployment claims while also effectively identifying and stopping excessive fraud. ODJFS considers ongoing investment in this project critical in its efforts to create a scalable, flexible, efficient, and effective UI program that can better serve citizens during normal times and withstand potential future crisis. At the same time, ODJFS continues to invest in its relationship with IOP to bring the lessons learned from this project to its other important human services programs.