Northwestern Mutual
Enterprise Software Modernization

Project Brief

Problem:

Many of Northwestern Mutual's internal applications were rapidly aging, accruing technical debt and increasingly unable to interface with newer products and their data.

Through business requirements and design thinking workshops, the primary goals were uncovered and decided: simplify and reduce, as follows:

• Address the product’s difficulty to create an easier, more efficient workflow for users
• Reduce reliance on separate platforms for data inquiry
• Utilize established engineering frameworks and design patterns to maximize the limited development resources

The primary goal from business partners was creating a unified database to further streamline output and data-driven insights for a more dynamic experience overall, thus guiding the effort to update RCS allowing the software to integrate more easily with other software across the company.

Text

Users described the application as:

• "Very slow to onboard"
• "Full of workarounds to make working with other software easier"

The catalog of festival films

By working directly users familiar with the platform and its workflow, I was able to directly test design hypotheses against feedback throughout the entire process, making minute adjustments in an agile approach.

Text

Text

big text big text big text big text big text big text big text big text big text big text big text big text big text big text big text big text big text big text big text big text big text big text big text big text

Text

Text

• Text
• Text
• Text

This led to a guiding question:

How might we big text big text big text big text big text big text big text big text

Project Details

The Replacement Compliance System, known internally as RCS, was legacy enterprise software used by the reinsurance division within the larger underwriting sector at Northwestern Mutual.

Design leadership recruited me and my team to lead the platform's redesign as part of an ongoing effort to modernize software throughout the company.

The primary goal requested by business partners was allowing RCS to integrate easily and share data with other software across the company, creating a unified database to further streamline output and data-driven insurance insights for a more dynamic experience overall.

Early in the project lifecycle, I spent a lot of time working to comprehend the intricate user flow a reinsurer would navigate through multiple programs and screens to complete a single work assignment. The challenge lay in understanding how users navigated multiple systems in sync: when, how, and why they accessed specific data points based on nuanced process flows.

This complexity was referred to throughout the project as the swivel chair effect, demonstrating how users constantly moved back and forth between separate applications to complete a single task.

Through my efforts to understand the primary workflow, one of my clearest goals from the outset became to simplify and reduce as much as possible into a unified design. RCS was not easy to understand or intuitive to use when I was first exposed to the product.

Many apparent issues had grown organically over time as the product's needs and scale increased. The standard reinsurance work assignment typically required three months of onboarding with RCS for new employees to master, which excluded any potential problem or edge cases outside the primary flow and a clean, standard work assignment..

Many apparent issues had grown organically over time as the needs and scale increased, such as the heavy reliance on learned behaviors. The text-based design created a very mechanical understanding of interaction in its users who learned the exact number of clicks and keystrokes needed to complete tasks.

RCS typically required three months of onboarding for new employees to master a standard work assignment due to the many workarounds in use. This learning period excluded grasping understanding of edge cases outside a clean, typical work assignment or any potential problems outside the primary flow.

These factors manufactured a work flow where none of the tactics felt natural for new users or flexible in cases of outlier work assignments. The overreliance on highly tuned muscle memory made onboarding protracted and painful. It also increased the chances of potential user error through the sheer difficulty as well as the complexity from the swivel chair effect.

A better design would require less memorization from users and provide more understanding at a glance when needed. These conclusions revealed a clear opportunity to optimize the experience based on design principles and user needs.

One of the biggest roadblocks during the exploration phase was the sheer amount of different data points users examined when selecting a standard work assignment to process. Within my team, I focused on reorganizing RCS's summary / work assignment pages. The legacy design displayed around fifteen data points presented in a table-cell format which required horizontal scrolling to examine thoroughly. The information lacked any hierarchy to help understand the relative importance of each data point. When inquiring about the importance of each piece of information, SMEs emphasized how each was required and vital to the process. Legal compliance placed many restrictions, which in turn affected my ideation process.

This limitation, the inability to remove or alter any data points due to legal standards and project SME’s deeming each data point critical, seemed like a significant setback at first. The challenge became how to design around this constraint. To uncover ideas on a path forward, I conducted informal user research – casual interviews, anonymous feedback, and informal polls – to learn how users interacted with each data point in the current design. Synthesizing research findings helped me understand not only what users needed but when and how frequently they needed it.

By understanding each data point’s relative importance, I updated the design to group points by related functions. I employed metadata to create visual hierarchy (through typographical sizing and subtle use of color), displaying correlations between related points and helping users more quickly scan the table to assess potential work assignments. This shift also imbued the table with some progressive disclosure, highlighting high-priority details while diminishing, though not outright removing, less critical data points.

This research-backed design shift created a quicker, more intuitive view of accessing and assigning work tasks for both general and power users. Its cleaner layout preserved the required data while eliminating the horizontally scrolling table's atypical, unintuitive design pattern, a clear byproduct of organic product growth.

Key ?? takeaways:

• ?
• ?
• ?
• ?

As a team, we reimagined the overall input experience to mimic the familiar layout and language of existing paper forms used by the reinsurers, tapping into the mental models already prevalent in their minds. Replacing the fragmented navigation could create a more unified, flexible experience. RCS’s legacy experience utilized a tabbed structure, segmenting inputs into several discrete groupings with distinct inputs within each tab. In its place, we implemented a seamless vertical scroll, allowing users to view and access the entire flow holistically.

This design shift reduced friction and the need to navigate among different sections. It reduced the users’ need to memorize where each text input existed. Users could navigate through their work assignments more freely, creating a more streamlined experience. Additionally, it eased new users into building mental models of the connected throughlines across their work assignments and required paper forms. This subtle tweak helped to eliminate hidden dependencies on mechanical muscle memory and homegrown keyboard shortcuts.

Beyond layout, we focused heavily on language. The original system used jargon that did not match our users’ mental models and was inconsistent with the copy used on the internal forms they used to complete their assignments. We rewrote system labels to either match the language of existing forms or clarify intent with additional help text where appropriate or necessary. This subtle shift helped reduce cognitive load, especially in instances of onboarding, and gave the entire interface a more approachable tone.

In parallel to design efforts I championed personally, an adjacent initiative helped to overhaul the correspondence flow users accessed within RCS to send / receive information from various internal and external parties. This functionality managed communication with customers and internal divisions when reinsurers needed additional input or clarification to complete work assignments. While I did not directly design this flow, I collaborated with my fellow designer through consistent early-stage feedback, helping align the proposed correspondence solution to the overall product vision and user needs. This feedback played a critical role in reinforcing the continuity of the reinsurance platform, further underscoring the team's vision of a modern, straightforward user experience.

Key ?? takeaways:

• ?
• ?
• ?
• ?

Outcomes

As the proposed platform moved from refinement to development handoff, its impact quickly became clear. The new design significantly improved users’ throughput rate. In prototype testing sessions, reinsurers could complete assignments more efficiently with fewer errors. The learning curve shortened dramatically due to clearer language, simplified flows, and reduced dependency on memorizing actions to offset quicks within the previous platform. This increase in ease of use also decreased the effort needed to onboard new users. Perhaps most impactfully, the swivel chair effect diminished as the updated design consolidated more data streams within the RCS platform.

From a technical standpoint, the modernized structure allowed engineering teams to build the new platform by reusing large portions of previously built code using standardized components. This resource reduction helped to maximize engineering effort in areas that needed additional development as well as enabling future integrations into the company-wide modernization with less development effort. This compounded benefits for our users: fewer external programs were required to complete work tasks and updates could roll out more reliably.

My work also laid the foundation for several Day Two enhancements, including a suite of administrative tools designed to empower administrators. These tools existed in the prior version of RCS, though they were confusing to all except the most experienced, senior users. Through continued research and informal interviews, I came to understand the general perception of these tools: “Well, I know they are there and I occasionally go to them to double-check information but they don’t really do much.”

This insight clarified a major opportunity. By investing a modest amount of design attention over the course of the project, I was able to build clearer understanding and surface untapped value in these tools. I came to recognize their value in providing agency to users of the RCS system, removing a need for backend support. These tools could be harnessed to give reinsurers the agency to amend their own databases of key product data, related companies, and required addresses. I imagined beyond current functionality, proposing new features such as a correspondence builder to draft basic boilerplate language, for which users had previously been using keyboard shortcuts as a workaround solution.

This suite of tools served to further support the project goals of a more efficient workflow and a reduction in the need for external support or software through greater user autonomy. While outside the original scope, these enhancements pushed the system further toward our shared goal: creating a more efficient platform that enabled users to work with greater ease and flexibility, all informed by research and feedback collected during the initial design phase. Administrators would be able to make changes easily and efficiently while general users can further increase their throughput by requiring less intervention from admins.

Key ?? takeaways:

• ?
• ?
• ?
• ?