Steered the direction of a startup’s legacy software redesign
Resale items are processed in warehouse facilities like this. Associate workstations (not pictured) are set up on the perimeter of the warehouse floor.
Introduction
This is the story of how I used UX discovery methods to explore and prioritize opportunities for the redesign of a startup’s legacy warehouse operations software.
Background
In the fall of 2022, Trove's existing software had served its purpose but was hitting its limits in terms of usability and performance. As the company aimed to scale to serve its next 50 brand partners – among the likes of Patagonia, REI, and lululemon – it became clear that the current warehouse operations software, originally designed by engineers and using custom flows for each partner, wouldn't suffice.
A complete reimagining of the offering was necessary: We needed to create a scalable SaaS product that could be used by warehouse operations teams to manage resale programs and process items efficiently.
Challenge
We didn’t know exactly how to translate what worked well from the old system into the new. We knew we wanted to streamline and standardize various operational workflows, as well as tie in a dynamic rules engine for intelligently routing items to their most profitable destination. How would we balance a standardized flow, applying the resale intelligence that Trove had gained over the years, with a flexible solution to meet the unique needs of each partner?
We needed to explore new, compelling features while ensuring we maintained scope to rebuild a faster, more intuitive product for operations teams to use. How would we balance the needs of administrators (who managed the operations) versus associates (who collectively processed thousands of items daily)?
And, given our lean startup environment with limited resources, the scope and timeline for this project would be tightly constrained.
Solution
Business context
Because I was still relatively new to Trove, having started there just 6 months earlier, I created an ecosystem map to ensure I understood the actors and relationships in our offering.
Sketches and note-taking in FigJam: I gathered feedback on my initial draft and informed my design process with executive documentation on the business case.
Generating insights
I was already familiar with the legacy system, but I re-familiarized myself with an end-to-end audit of the flow, to ensure I was aware of small details or hidden features. I decided on elements to keep in the new solution, including a stepper component for giving the user context. Others, like the color system, needed to be iterated on, as the legacy system lacked a clear visual hierarchy and did not have high enough contrast levels for accessibility (by the WCAG 2.0 AA standard).
I took screenshots and notes and recorded Loom videos as I walked through competitive solutions, taking inspiration from comparable user experiences, such as Apple Trade In, Typeform, and Airtable. Keeping organized with these materials enabled me to quickly call back to my references and loop stakeholders into my sources.
User needs
I created detailed personas for our two main user groups: the “Warehouse General Manager” and “Warehouse Associate”.
From my observations, I also learned that users were required to use external job aids (e.g. laminated reference sheets, binders, and browser tabs linked to training materials Google Slides) to accomplish their tasks in the software. For each brand partner, the Warehouse Associate would have to be trained on that brand’s products and resale eligibility rules. For example, REI accepted items that were in “Grade D” condition, but lululemon and Levi’s only accepted “Grade B” and above. Associates had to be trained on each brand and program, and then reference back to their job aids to make decisions. We needed to build this training guidance into the software, to reduce overhead for training and minimize context-switching for associates.
Key insights included:
The need for on-screen instructions to reduce dependency on physical job aids.
The importance of visual guidance, especially for ESL (English as a Second Language) users.
The value of a company KPI-driven "Save a click" design principle to enhance efficiency.
A persona card that I created to help the team empathize with our Warehouse Administrator users, tapping into their emotional needs and surfacing their business objectives
Feature prioritization
By using the Kano model of plotting features on a Satisfaction-Functionality axis, I visualized the features we knew we wanted, along with the ones I had discovered as part of my competitive review.
While my ideas to create an onboarding flow and reporting dashboard were well-received, unfortunately the timeline was just too short and small to explore these features. For the MVP solution, internal Trove staff would support new user onboardings and continue surfacing insights in Looker.
An inside look at my FigJam workspace for prioritizing features, defining research questions, and collaborating with the lead product manager. The “Rhino” and “Zebra” became our “Warehouse General Manager” and “Warehouse Associate” personas.
Prototyping
I drafted a set of screens in FigJam, using simple shapes and text blocks to create low-fidelity wireframes. I shared these early concepts with stakeholders to get feedback before allocating my time to higher-fidelity prototypes.
After aligning on basic content and layout structure, I developed a functional prototype in Figma to test with users.
The Figma prototypes of key user flows
Testing
I wrote a test script to assess how well the prototype met user needs, asking both qualitative and quantitative questions to measure success.
The flows earned high marks on their intuitiveness and minimalist, high-contrast design; quantitatively, users virtually all succeeded with 0 errors on their tasks, meeting my goal of an 80% task success rate.
Admin users appreciated the ability to freely configure program rules in a user interface, whereas the rules were previously embedded in the code and needed to be updated by engineers. They also responded positively to the Preview feature I came up with, which enabled them to visualize what the associate would see as they processed items.
Most participants gave feedback that the terminology we were using for an admin feature was confusing. I made a recommendation to the lead product manager to replace the term “criterion” with “variables”, to simplify it a bit without losing the intended meaning.
A few pages from the research script I created and used to assess my prototype
Throughout the process, I remained flexible and open to feedback, involving stakeholders from design, engineering, product management, and operations. This approach allowed me to quickly iterate on my prototype and improve my understanding of our business.
Unfortunately, leadership put the project on hold, and I paused my design work at this point.
Results
In just 4 weeks, I achieved three key outcomes:
Persona definition: Clarified the two primary user personas to serve with our MVP. Identified important secondary personas to explore further post-MVP.
Feature prioritization: Aligned on a prioritized feature list for the MVP with key stakeholders.
Prototype usability: Wrote a metrics-driven research script and implemented research-backed improvements.
Kailyn excels in the rapid iteration of designs, always ensuring that our users remain at the forefront of every decision. Her approach to design is both thoughtful and strategic, as she consistently asks insightful questions that reveal the underlying needs of our users. This user-centric mindset has been pivotal in creating intuitive and user-friendly designs that resonate deeply with our audience.
Branko Raic
Lead Product Manager @ Trove
Next steps
At the time, my next steps would have been to:
Finalize the initial set of components in our custom design system
Create implementation notes and do a warm handoff to the engineering team
Steward the MVP designs through successful implementation and UAT
We did end up picking this project back up the following spring. The foundation I laid with my design work during this discovery project provided the strategic direction that gave us a strong foundation when the company was ready to restart.
Critique
In the spirit of continuous improvement, here’s what I would do differently next time:
Quantitative data: While qualitative insights were rich, incorporating more quantitative data from the legacy system (e.g., processing times, error rates) could have provided a stronger baseline for measuring improvements.
Timeline management: The ambitious scope of the project for a 4-week timeline meant that some potentially high-value features (like onboarding flows and reporting dashboards) had to be deprioritized. In future projects, advocating for a longer discovery phase or a phased approach could allow for the exploration needed.
Continuity planning: Given that the project was put on hold, developing a more robust handoff plan or documentation strategy could have ensured smoother continuation if/when the project resumed.
