Online Planning Experience: Expense Categories

Many Americans have a retirement goal, but with the financial landscape in the U.S. becoming increasingly complex, retirement planning has become more critical than ever before. Even with greater access to 401(k)-style plans, many people still aren’t fully prepared. In fact, according to the National Council on Aging, 80% of older adults are either struggling financially or at risk of economic insecurity in retirement.

That’s where having a financial advisor can make a difference—helping individuals build a plan that navigates the challenges of retirement planning. Morgan Stanley goes a step further with a responsive web tool called OPE (Online Planning Experience), which makes it easy for clients to provide retirement goal details.

To view this project as a presentation, click here.

Overview

The tool helps users outline their retirement spending plans, savings accounts, and expected income sources. However, when the tool asks for pre- and post-retirement expense details, the categorization of the expenses appeared confusing and misleading.

In this 6-month project (combined with other projects for the same release), I was the lead designer and researcher, collaborating with business, engineering, financial advisors, design language, UX, research, and stakeholders to drive the restructuring of the expense categorization experience, from research to release.

Project Goal

Since OPE sends the user’s details to GPS (Goals Planning System) – an internal tool used by the financial advisors, the information needs to be accurate so that the financial advisors can build the right strategy to help clients meet their retirement goals. Not only does the accuracy help financial advisors and our users with the planning, but it helps business teams keep accurate account of how expenses are being categorized across plans.

In this project, we want to identify what is the ideal experience, so that clients can intuitively find and navigate to the right destination when adding details of each type of expense. We want to ensure that the FA (financial advisor) and client can collaborate smoothly, avoid miscommunications, eliminate confusion, and have the same understanding around expense categories across both parties. Also, we want to maintain consistency in data between OPE and GPS.

The Process

While this project was heavily focused on information architecture and user research, the process was still similar to many of my other UX projects. I dove into the project gathering all the information I can to understand the problem, challenges, technical constraints, the goals and needs of everyone involved in the experience, explored a range of solutions that address the pain points, tested the solutions, then bringing the solutions into production.

Job Story

To help think about the context and user’s motivation, let’s use this job story as a starting point:

When I am adding a new expense, I want to be able to quickly find and navigate to the correct category when documenting each expense, so that I can provide the most accurate information to my financial advisor.

Understand: Discovery and Analysis

When designing the screens for OPE, we are generally required to align the experience with GPS, so that financial advisors can easily decipher the data and create the right financial plan for clients. However, sometimes what is intuitive to financial advisors, might not always be intuitive to our users.

For context, in the OPE experience, the user arrives at the list of categories through a couple different paths from the dashboard. They can either go through Retirement Spending, or Other Expenses section.

At the start of this project, OPE used these categories:

In comparison, GPS used these categories:

Both experiences contain areas of confusion and poses a number of questions.

In regards to GPS’s version:
• If bought a used car, would I enter it into “New Car”, because it’s a new car for myself, or is it a “Major Purchase”?
• How new does the car need to be, in order to be categorized as “New Car”?
• If I’m renovating a new home, would I enter it into “New Home” or “Home Renovation”? Is there a difference, and how would it affect the planning outcome?
• If the Other Expenses section is not for day-to-day expenses, why is Credit Card supposed to be included here?
• Could mortgage be categorized under “New Home”, instead of “Loans”?
• Why was “New Home” and “Home Renovation” broken out into separate categories?
• Is Other Expenses section for capturing all expenses, before and during retirement, or is it just during retirement?

As for the OPE version:
• What order are these categories following? Why does OPE start with Education?
• How is this screen supposed to be read? From top to bottom, left to right? Or left to right, top to bottom?
• Where would loans be entered into?
• Why was “Celebration” renamed to “Weddings”, when it includes celebratory events like retirement, milestone anniversary, or birthdays?
• If I’m renovating a home, would I enter it into “Renovation” or “Home”?
• Why was the help drawer removed? (The help drawer contained supplemental information around the category definitions)

To get a better understanding behind how expenses were being recorded on the receiving end, and identify any constraints we had, we spoke with the financial advisor team that uses the GPS tool. Currently, the GPS uses these categories and definitions:

The challenges

Based on discussion with the financial advisor team, we’ve identified a couple more challenges or constraints:

1. All user-generated data from the OPE side will be sent to GPS, which is handled by a separate team.

2. The current taxonomy and nomenclature used in the GPS system is what FAs are comfortable with and makes the most sense to them. Regardless of the research findings we receive around the user’s mental model, we won’t be able to change what is currently in GPS. Even if we were able to change what is in GPS, the technical effort would be too large to be considered.
   
   So, how might we restructure the IA of the categories so that it is intuitive for users, and scalable in a way that doesn’t increase too much scope, but also map the data to a completely different system that might not match the user’s mental model, while adhering to the design language used throughout the cross-platform product, and meet accessibility guidelines?

Card Sort Study

To see how users performed with GPS’s structure, I proposed conducting an unmoderated, hybrid card sort study using Optimal Workshop, and gather additional qualitative feedback through UserTesting. While our team hasn’t used Optimal Workshop before, my experience with it in past research projects has shown that its analysis features are more robust than those offered by UserTesting.

To move forward, I connected with multiple team members to identify the right contact at Optimal Workshop, ensuring I could secure a temporary account while adhering to our firm’s security protocols.

Research Goals

The goal of this research study is to understand if Morgan Stanley look-alike users can navigate the experience of documenting their different types of expenses. To determine this, the study should:
• Identify any confusing taxonomy or nomenclature
• Identify groupings and naming conventions that users understand more easily

Research Methodology

What was tested

Based on the category names and the definitions provided by GPS, I worked with the content strategist to produce the copy used for the card sorting test on Optimal Workshop. A few extra cards were added to provide more context, and also for research purposes to learn more about the users and their mental models.

Key takeaways

1. According to the research findings from the card sort study, we found that users bucketed these content pieces into a wide array of categories which did not match the GPS categories:

2. Users wanted more context for each category and content piece. With more context, users would have a stronger idea on where to sort the following cards:

* Some cards, not originally part of GPS’s list of example expenses, were added to provide more context, and also for user research purposes to learn more about their mental models

3. Based on written responses or verbal feedback, users felt that some of the categories could be consolidated more, especially around home and auto. Users agreed with keeping all loan-related content pieces under Loans.

4. Users had some confusion around the category names. Some are too similar in meaning to each other (Gifting vs. Celebration; Home Renovation vs. New Home) and some do not match the user’s mental model altogether (Travel, Celebration). Other categories were too granular (New Home, New Car).

5. The content under the “Other” category had the lowest agreement among users. The results were vastly different from one user to another.

Recommendations based on card sort findings

→ To view the full card sort study with detailed insights, please click here.

Explore: Ideate, Design and Test, Analyze

Tree/Treejack Test

Now that we’ve gathered sufficient amount of insights and learnings from the card sort study, we are ready for the next part of testing – the tree test.

Research Goals

In this research project, I conducted a tree test with 3 versions – GPS’s version for benchmarking purposes, and 2 proposed structures. This tree test will evaluate the proposed structures, and determine which one should be implemented. If there are areas that still need revision, this tree test will also provide insights on how to improve those areas as well.

Research Methodology

What was tested

Taking what we learned from the card sort study and using the recommendations as guideline, I collaborated with the content strategist to explore new structures that address the pain points found in the previous test. We concluded with 2 potential structures that could improve the experience.

How the tree test was conducted

Users were given specific tasks, and then asked to find the location of the item within the structure. This essentially evaluates how well your information architecture performs in a real-world scenario.

Example task:
Someone in your family is having a wedding in the future, and you are paying all costs related to this event. Where would you enter this expense?

Key takeaways

1. Test A and B both consistently outperformed the original structure

2. Having items exist in more than one place in the structure, consolidating the categories, and renaming the categories, and proved to be successful overall

3. Between Test A and B, Test A was more successful

• Test A had more polyhierarchical items than Test B (ie. Car, Home)
• Only 1 part of the structure had more than 1 level (Major Purchases/Expenses)
• No “Other” category

4. Although Test A performed the best of the 3, the structure can be improved

• Some users still had trouble finding Purchase of Land
• While most users found the right destination, a few expected to find it under Home or Buying a Home

5. The Loans category might still be confusing to some

Recommendations based on tree test findings

For immediate next steps, I made the following recommendations:
• Add Purchase of Land under Home, in addition to having it under Major Purchase/Expense category.
• Based on research studies, use alphabetical order for the list

Since there were some remaining questions around improving the IA, I suggested that we run another round of tree tests in the future to identify if:
• Loans is a necessary category
• Users prefer the categories listed in alphabetical order, or in order of most commonly used categories (to confirm the other research studies found on this topic)

→ To view the full tree test study with detailed insights, click here.

Materialize: Implement

The expense category experience is available across all platforms, built in responsive web format. I documented the full userflow–from the dashboard to where the user enters all details about the expense. I also compiled the UI specs–while collaborating closely with business, tech, stakeholders, financial advisors, accessibility, and UX–detailing the back-end mapping of the categories and subcategories to the GPS tool.

Once all specs were complete, I worked closely with the dev team to implement the designs, and ensure that the final product matches the intended experience before releasing this to the public.

Final Designs

What I learned from this project

1. Who knew that restructuring expense categories could be such a complex project! From uncovering how the different systems needed to communicate with each other, to understanding not just the user goals/needs, but also the financial advisor’s role in the user journey, all the tech capabilities and limitations, and paving the first path for the team to use Optimal Workshop for card sorting and tree tests–was a tremendous opportunity to drive the project from many different angles while collaborating with a significant number of team members.

2. Despite testing multiple times, there’s always room for more improvement. As mentioned at the end of the tree testing, there were a couple more open questions in regards to how to make the experience even better.

Thanks for reading, and I hope this was as interesting and enjoyable for you as it was for me!