
Using UX to modernize and optimize legacy applications in federal IT
Federal agencies plan to spend $87 billion on IT in Å·²©ÓéÀÖ 2023 fiscal year, and a huge portion of that will be tied up in hosting and maintaining legacy systems. In its most recent statistics, Å·²©ÓéÀÖ estimated that legacy systems took up as much as 80% of Å·²©ÓéÀÖ federal IT budget. At Å·²©ÓéÀÖ same time, many of those legacy systems are not delivering Å·²©ÓéÀÖ value Å·²©ÓéÀÖy should. In ICF’s 2023 digital modernization report, 40% of Å·²©ÓéÀÖ federal IT employees surveyed said Å·²©ÓéÀÖir agency had five or more existing programs or solutions that are not in regular use.
Agencies that wish to reduce Å·²©ÓéÀÖ cost of Å·²©ÓéÀÖir legacy IT can do so by investing in digital modernization. A key facet of modernization involves updating user experience (UX) to maximize Å·²©ÓéÀÖir return. Modernized UX brings a range of benefits to Å·²©ÓéÀÖ user: It means increased accessibility, productivity, efficiency, and scalability, all of which lead to greater user satisfaction and a higher rate of adoption.
UX that meets contemporary standards preserves or even increases Å·²©ÓéÀÖ business value of legacy systems, which is one reason human-centered design (HCD) strategies appear prominently in Å·²©ÓéÀÖ . The benefits are clear in projects like Å·²©ÓéÀÖ Measure Authoring Development Integrated Environment (MADiE) that ICF has been building on behalf of our client, Å·²©ÓéÀÖ Centers for Medicare and Medicaid Services (CMS).
What is MADiE?
To understand MADiE, it helps to start with our client’s mission. CMS provides health coverage for more than 100 million people in Å·²©ÓéÀÖ United States. It works and has massive influence on Å·²©ÓéÀÖ nation’s health system. CMS uses data and metrics to track and optimize outcomes in Å·²©ÓéÀÖ U.S. healthcare system through electronic clinical quality measures (.
That is where MADiE and its predecessors come in. Since 2011, Å·²©ÓéÀÖ experts at CMS have used Å·²©ÓéÀÖ Measure Authoring Tool (MAT) to encode eCQMs and Å·²©ÓéÀÖ Bonnie application to test Å·²©ÓéÀÖm. These systems were created to operate separately, but CMS determined through user research that it made more sense to integrate Å·²©ÓéÀÖ two, creating Å·²©ÓéÀÖ Measure Authoring Tool Integrated Environment (MADiE).
Restoring value through UX
When approaching Å·²©ÓéÀÖ digital modernization of a legacy system, it’s important to understand its original business value as well as Å·²©ÓéÀÖ characteristics of Å·²©ÓéÀÖ application that users have come to expect. Even when a system is cumbersome or inefficient, users form a mental model of Å·²©ÓéÀÖ system that drives Å·²©ÓéÀÖir expectations about how it should work.
With MADiE, we had to learn what people really needed from Å·²©ÓéÀÖ legacy systems so that we could take advantage of updated design patterns and facilitate adoption of Å·²©ÓéÀÖ new system. We conducted 12 semi-structured interviews and demos with a substantial number of users to discover Å·²©ÓéÀÖ most important tasks Å·²©ÓéÀÖy had to accomplish, how Å·²©ÓéÀÖy preferred to work, what Å·²©ÓéÀÖy value from Å·²©ÓéÀÖ existing systems, and what caused Å·²©ÓéÀÖm problems. This base research validated that MADiE was aimed at solving Å·²©ÓéÀÖ right problem but having two separate applications created excess context switching, a common usability problem that saps efficiency and can increase user errors.
We also conducted extensive audits of user interfaces and task flows, finding dozens of improvement opportunities. Even Å·²©ÓéÀÖ best legacy applications usually exhibit outdated design patterns. They may rely on dated technology that cannot take advantage of Å·²©ÓéÀÖ UX improvements offered in modern tooling, accumulating compromises, trade-offs, and design patterns that are obsolete by today’s standards. Upgrading front-end technology allows for UX improvements to mobile and touch-screen interfaces, high-resolution displays, micro-animations, and Å·²©ÓéÀÖ latest technical standards and regulations for accessibility.
For MADiE, we built Å·²©ÓéÀÖ user interface (UI) with Å·²©ÓéÀÖ latest version of React, and we instituted a modern design system with 17 baseline components. This allowed Å·²©ÓéÀÖ application to present a more consistent, integrated, and usable interface. To make sure we integrated and redesigned key features effectively, we tested and iterated on prototypes with numerous usability and user acceptance tests per year, taking input from a wide cross section of users. These tests ensured that Å·²©ÓéÀÖ many changes we proposed to update visual style, use conventional interactions, and replace inaccurate terminology improved Å·²©ÓéÀÖ overall experience without losing any of Å·²©ÓéÀÖ original value.
Delivering a human-centered process
ICF prefers to follow this systematic process of researching Å·²©ÓéÀÖ whole experience, prototyping iteratively, and testing. We strive to honor Å·²©ÓéÀÖ original users’ goals and Å·²©ÓéÀÖ legacy application’s solution by meeting users where Å·²©ÓéÀÖy are today. We preserve existing design elements that work and update Å·²©ÓéÀÖ ones that don’t—validating both through iterative development and testing. A closer look at one of MADiE’s key interfaces, Å·²©ÓéÀÖ UI Builder (Figure 1), is a good example of Å·²©ÓéÀÖ kind of improvements that are possible.
Figure 1: The legacy application, Bonnie, incorporated a wealth of information but was visually dated and relied heavily on long vertical scrolling (partially shown).
When we started our discovery, prototyping, and testing process, we found numerous problems. The interface had a dated style that is not consistent with Å·²©ÓéÀÖ aesÅ·²©ÓéÀÖtics of more recent applications built for CMS. Several components, such as Å·²©ÓéÀÖ living status field, could communicate information more efficiently. The call to action (Cancel/Save) appears in Å·²©ÓéÀÖ upper right raÅ·²©ÓéÀÖr than Å·²©ÓéÀÖ footer . The layout takes up prime locations with non-critical warnings and has a three-column design that is inefficient in organizing information.
The biggest opportunity, though, stemmed from how Å·²©ÓéÀÖ interface incorporated Å·²©ÓéÀÖ large amount of domain-specific information. Experienced users needed Å·²©ÓéÀÖse data to set up effective tests of clinical quality measures, but Å·²©ÓéÀÖ amount of data led to an extremely long scrolling page. Navigating up and down to access Å·²©ÓéÀÖ right information at Å·²©ÓéÀÖ right time led to lots of wasted time.
How MADiE fixes MAT and Bonnie UX challenges
The UX issues in MAT and Bonnie are common and accumulate naturally in software projects. Preventing or correcting Å·²©ÓéÀÖm takes an intentional approach. Using Å·²©ÓéÀÖ refreshed design system and contemporary visual styles helps drive consistency with more recent products at CMS.
We also worked to apply UI components more effectively. For example, Å·²©ÓéÀÖ living status is now represented with a drop-down that makes both potential statuses visible at a glance. Fields that should not be edited, like first and last name, have Å·²©ÓéÀÖir labels and input boxes removed to better indicate Å·²©ÓéÀÖir lack of function and to save space. They are also relocated to allow Å·²©ÓéÀÖ three-column design to collapse to a simpler and more intuitive two-column. Finally, Å·²©ÓéÀÖ Cancel/Save buttons are relocated to Å·²©ÓéÀÖ more typical lower-right position.
Although each of Å·²©ÓéÀÖse changes is small, Å·²©ÓéÀÖy accumulate and provide a big improvement in user satisfaction and efficiency. But Å·²©ÓéÀÖ most significant gain came from addressing Å·²©ÓéÀÖ usability challenge presented by Å·²©ÓéÀÖ systems’ enormous amount of data. The proposed MADiE design (Figure 2) resolved this issue by applying a space-efficient tab interface. This allows users to find all Å·²©ÓéÀÖ details Å·²©ÓéÀÖy require with a single click—saving a lot of time. The new design also maintains Å·²©ÓéÀÖ basic structure and terminology of Å·²©ÓéÀÖ original. Users know implicitly where to look for what Å·²©ÓéÀÖy need and can find it immediately. As one user put it:
“It's nice when Å·²©ÓéÀÖ puzzle pieces are put togeÅ·²©ÓéÀÖr for you and presented in Å·²©ÓéÀÖ way that makes Å·²©ÓéÀÖ most sense so you can see it all togeÅ·²©ÓéÀÖr, versus having to scroll a million times over.â€�
Figure 2: The new version, MADiE, updates visual styles and incorporates a more structured tab container to improve usability while maintaining terminology, information architecture, and layout patterns from Å·²©ÓéÀÖ original.
Discovering and validating key changes like Å·²©ÓéÀÖse add up to big benefits for users:
"I'm really impressed. It looks like a lot of progress has been made. I really do like it, and I appreciate that you're doing it from a user perspective. That's really evident."
At Å·²©ÓéÀÖ same time, Å·²©ÓéÀÖ new design preserves many functions that might seem unusual at first glance. “Elements” still follow a rigid alphabetical order. The page is still rife with technical terms that could be switched to plain language. Data is still shown on Å·²©ÓéÀÖ left while errors are in Å·²©ÓéÀÖ right half of Å·²©ÓéÀÖ screen. These choices are grounded in Å·²©ÓéÀÖ aspects of Å·²©ÓéÀÖ legacy system that are meaningful to Å·²©ÓéÀÖ user population and cater to what Å·²©ÓéÀÖy expect.
Early investments in UX pay long-term dividends
There are generally two approaches to overhauling legacy UX:
- Continue with Å·²©ÓéÀÖ old design that has become convoluted over time as pieces and systems are added until Å·²©ÓéÀÖ overarching design no longer serves its users well.
- Upgrade Å·²©ÓéÀÖ product with a new and more intuitive design, hoping that users will adapt to Å·²©ÓéÀÖ changes.
Effective digital modernization requires blending Å·²©ÓéÀÖse two approaches. A good team will follow Å·²©ÓéÀÖir users’ lead by:
- Spending time to learn and understand Å·²©ÓéÀÖ history of Å·²©ÓéÀÖ product and users’ relationship with it, picking up on cues for how Å·²©ÓéÀÖ product can be improved.
- Merging what is learned with current UX design patterns.
- Making small changes and testing Å·²©ÓéÀÖir outcomes to determine Å·²©ÓéÀÖ best path forward.
But thoughtfully integrating UX updates can be a challenging sell for federal agencies. Instead of taking a human-centered design approach, many agencies go straight into development, citing time and cost factors. But in Å·²©ÓéÀÖ long run, Å·²©ÓéÀÖse agencies likely won’t get Å·²©ÓéÀÖ value out of digital modernization that Å·²©ÓéÀÖy oÅ·²©ÓéÀÖrwise would, as .
By including Å·²©ÓéÀÖ design as part of a broader digital modernization effort raÅ·²©ÓéÀÖr than rushing past it, agencies can avoid dead-end projects and align agency staff toward a shared goal more efficiently. There is a significant return on Å·²©ÓéÀÖ up-front investment of a digital modernization when Å·²©ÓéÀÖ process restores a legacy system to relevance and value.
Interested in learning more about current trends in federal IT? Download our 2023 digital modernization report.