Modernizing unemployment insurance: Truss leads the way
Truss is a leader in UI Modernization. There. We said it. Our recent success with the New Jersey Department of Labor (NJDOL)—on behalf of the US Department of Labor (USDOL)—is something we can’t stay quiet about any longer.
I led design for most of the project, working with a cast of excellent fellow designers including Liz Lin (the initial lead), Gaby DiSarli, Madison Ross-Ryan, and Kate Saul. Kate recently left Truss to co-create the USDOL’s own content design/strategy practice! This project alone has been about half of my overall time at Truss, and I’ve witnessed multiple instances where our values have shone through. I’ll share those throughout a few forthcoming posts, but for this first one, let’s just set the scene.
Our work with the NJDOL/USDOL began with many common problems that are associated with antiquated systems.
The claimant’s process experience was slow, clunky, and unreliable. Multiply that pain by the 430,254 claims submitted in 2022, or even worse in 2020 when the pandemic led to 2,041,115 claims that year!
Prior attempts to respond to the situation had fallen short and caused further discouragement. They needed a clear path to manage this key aspect of providing billions of dollars in benefits each year.
The agency was dealing with increasingly unhappy claimants and spiraling staff impact (cost / morale). An adequate solution was seemingly out of reach.
The struggle to modernize and innovate in civic service delivery to better meet operational, policy, and personnel needs.
In philosophy and in practice, Truss is a fiercely Agile development firm so in response to the problems NJDOL was experiencing, it was obvious implementing agile methodology meant redefining processes that were tightly coupled with legacy systems. For NJDOL, we delivered value in three critical areas:
🏗️ Infrastructure / Platform
Truss delivered results that are adaptable, scalable, resilient, and secure; built for today and for the future.➰ Agile Business Process
Truss delivered results that are collaborative, flexible, and efficient in response to dynamic situations and claimant needs.📄 Claimant Intake Application
Truss delivered results that are modern, flexible, and extendable; built with a high level of test coverage to enshrine intended behavior.
The end result? Truss vastly improved the claimant experience for unemployment insurance (UI) and improved NJDOL’s ability to respond to claimants’ changing needs. In the process, we set the conditions to reduce operational overhead: fewer agent interventions, a much more serviceable technical platform, and the organizational capacity to manage and evolve the system when Truss has signed off.
Working with workarounds
A survey of states early in this project found that many had undertaken modernization projects on various levels of their technology stack, from long-term storage mainframes to claimant-facing applications. Many of those states had paused, restarted, or pivoted over time, while others successfully completed them, and of course policies and the people who became experts in administering them have changed over time as well. With that in mind, we can see the reason the initial assumptions for this project were not fully on the mark, where we attempted building a single application, or “pull” model, for bringing states into the US DOL’s technical infrastructure. The pivot to a “push” model of creating a repeatable process and cloneable code allowed us to better work with these differing state-specific technologies, protocols, and knowledge.
Decades of these mismatched and inflexible software changes have led to technical workarounds—in both tech and human protocols—to record claimants’ truth as data. The experts we worked with in the DOLs of the US, New Jersey, Arkansas, and other states showed how they have been able to work with this fragmentation to ensure their constituents receive the due process and the money that they deserve. However, on many occasions this was in tension with our directive to modernize.
The most common way this manifested in our work on the intake form is as instructions to claimants to enter a certain value if a required question does not exactly match a claimant’s conditions. For example, if a currently disabled claimant in New Jersey starts a form and needs to answer a question about their disability recovery date, they would not be able to answer it as-is, since they have not yet recovered. Instead, helper text instructs them to enter the current date. When they submit the application, that date would then be on or after the “date of claim”, which is calculated as the Sunday prior to the submission date, and NJDOL agents know how to handle that from the following systems.
It’s worth noting that to reach this situation, the claimant must (inaccurately) answer “no” to an earlier question of whether they’re currently disabled, which would direct them to the Temporary Disability Insurance site instead of UI. However, since claimants must legally be able to submit this form regardless, they may anyway proceed to this step. A future iteration may seek a solution that addresses both front- and back-end systems, or it may dynamically change this question’s text to be based on their earlier “are you currently disabled” question. Either way, we were unable to reach a clear solution within our scope of work and time available, so this workaround persisted.
This is just one example of many workarounds that enable agents to work with data from systems meant to be fairly static, like a printed paper form, that might not easily yield the many different conditions claimants are in. Modernization—as a mission and set of success metrics—needs to reconcile this tension, while also allowing for continuation of service. While we may be able to suggest a more dynamic (vs static) way of asking questions like the above, it may require downstream changes to systems to accept new response types. Those changes may not be possible with the current tech or simply not ready for change with the limited capacity of a state IT office. When modernization projects are scoped, we must be ready to look at how both strategic and tactical goals affect, and are affected by, external systems.
How did we achieve it?
First, as strategic consultants we should be clear this was a limited work period and scope, and it builds on years of knowledge shared by our colleagues at USDOL, New Jersey and Arkansas DOL, advocacy organizations, and many individuals who went through the UI system as claimants and wanted to share their stories with us. We humbly thank them for their time to help us build. We also recommend pairing this case study with more comprehensive resources like New America’s Playbook to get the broader context.
The next few posts cover both the why and how of continuous development practices (CI/CD), agile and iterative culture, and user experience (UX) design, all delivered in a way that meshed with how our state and federal counterparts want to work: