From assessment to action
- My role
- Product design lead (sole designer, also driving product decisions)
- Team
- 1 product designer, 1 front-end lead, engineering
- Timeline
- Dec 2025 – Mar 2026
Description
Reframing an enterprise software-quality dashboard so users can decide where to act, not just see what exists.
Read case study
An enterprise software-quality dashboard mapped a user’s entire portfolio in a single view, but it only ever showed the current state, so it couldn’t help anyone decide where to act. I redesigned it from a static assessment into a trend- and activity-driven decision tool.
I was the only UX designer on this topic, without a dedicated product owner, so I worked closely with the front-end feature lead. My work covered research, strategy, design and roll-out planning.
The problem
Imagine a software architect with 100 systems in their portfolio. They open the product on a Monday morning and see a treemap of everything. The first reaction is “wow, my whole portfolio, mapped out.” It feels like being in control.
Over time, that changes. You know where each system sits on the map, and it never moves, because the treemap always shows the aggregated rating, and ratings change slowly. The view stops reflecting activity. Worse, the systems a team touched last sprint are often tiny blocks, buried. The most active systems were the least visible. The overview had become a giant navigation menu, and in interviews, users and consultants told me they mostly ignored it.
Two things needed fixing:
- Accessibility. The dashboard was built for people who already knew the product. Someone who hadn’t spent hours learning it would see a wall of data and close the tab. To scale beyond technical specialists, to managers and executives, the product had to be more opinionated: tell people what is good and bad, so they get a direct takeaway instead of interpreting raw data.
- From “what is?” to “so what?” and “now what?” The dashboard described a state. It needed to support a decision.
| What architects experienced | What we designed for |
|---|---|
| “Wow, cool map” | “Here’s what I need to do this week” |
| Which systems exist? | Which systems need attention right now? |
| I open it when I have time | I open it before I plan my week |
| A product I check | A product I rely on |
Research
The product grew out of consulting: it was designed around what consultants understood customers wanted, in a report-driven mindset. We delivered exactly what was asked for and kept adding features, without asking what underlying problem sat behind each request, or how often it would be used. The dashboard that resulted was static, like a slide deck: a fine report to review quarterly against a screenshot from last quarter, but not much help to someone deciding what to work on today.
I ran two kinds of research.
Two cross-functional workshops with consultants, customer success, front-end engineers and the product lead (1.5 hours each). The goal was one primary narrative and workflow for the epic, and a shared view of what to ship first given limited development bandwidth.
- We aligned on the problem, picked one primary persona and one trigger scenario.
- We mapped a high-level user journey (discover, decide, use, follow up), marking which stories were essential for a coherent first version.
- A “story, question, information” exercise: for each user story, what question is the architect really asking, and what must the overview show to answer it fast?
- A closing exercise, “what must I see in 10 seconds?”, defined the first screenful.
Three external validation sessions with tech leads and architects from three different customer companies. The hard part was that “which system is most urgent?” differs by company. To make it concrete, I put pairs of systems with different kinds and numbers of vulnerabilities in front of them, three pairs in total, and asked them to choose which to fix first. Forcing a decision, and listening to them think aloud about what else they would need to see, gave us something we had never had: direct user evidence instead of second-hand anecdotes from consultants. It also left the product team with documented insight beyond this project.
Decisions and constraints
Focusing on trend, without losing the aggregate. Shifting the focus fully to change over time was a hard call. People dislike it when they can’t find a thing where it used to be, especially mid-transition. We kept the aggregated view in a less prominent place, left the classic ratings tab as the default for new users, and rolled the redesign out incrementally over three sprints. Nothing was taken away. We added on top.
Pushing back on an “overall rating.” Early on we hoped to produce a single rating that said whether the portfolio is good. I came to see that it can’t work. Finding-based capabilities, like security, depend on the number of critical findings: you need to act whatever the rating says. Chronic capabilities, like maintainability, are long-term things you maintain and check less often. Merging them into one number would be biased and wouldn’t represent what’s going on. So I pushed back and emphasized different metrics depending on the kind of capability a page covers, instead of highlighting ratings everywhere.
A fixed set of periods. The period selector offers fixed options by default, which keeps data loading fast.
The solution
The first thing we added was a period selector. It looks like a small change, but it unlocks everything else: the dashboard no longer shows a state, it shows a delta. It’s built around three questions an architect asks on a Monday morning.
Where should I look first? Four KPI cards at the top give a health check in seconds: quality trend, acute risks, improving versus declining systems, and objective progress. We also removed repeated capability pages, so everything needed to compare across tabs lives in one place.
What’s changing, and is it getting better or worse? The hotspots view sorts systems by change: the most active ones are the biggest blocks, and color shows whether the activity made things better or worse. Big and red means a lot happened and it went the wrong way. The color scale is dynamic and based on the selected period, so you can compare this week to last. The system list can also be sorted by change, not only by rating.
Which systems need more attention than others? The big red blocks at the top of the list. No digging required.
Tabbed saved views
The treemap had four drop-downs that needed configuring to get the view you wanted. Interviews showed nobody used them, and each customer wants a different view as the best picture of their portfolio’s quality. So I turned that configuration into presets: tabs. Set up a view once, name it (“hotspots by team”, “objective progress”), and return to that shortcut every time. Customers can decide for themselves, or consultants can suggest views for their situation. For a SaaS product, surfacing deep configuration that users never knew would help them was a big improvement.
The tabs shipped as:
- Prioritize: systems with the most significant quality change and development effort.
- Objective progress: target achievement and volume changes, grouped by team or policy.
- Ratings: the classic risk dashboard showing the current state.
Outcome
- Shipped in March 2026. After feedback, the aggregated treemap tab was restored as the default in May.
- Engagement with a key diagram that had been used for a long time rose by more than 50%.
- In follow-up research, the same users from the original study said the change-based view moved them a concrete step toward a decision: from “which systems exist” to “which ones do I act on”. One described it as an easy way to get an action list of systems.
- The roll-out was smooth, with no dedicated product owner and incremental release across three sprints.
The ambition is simple: the product becomes the first tab an architect opens on Monday morning, not because a manager said so, but because they can’t decide where to focus without it. That’s not a feature, it’s a habit.
Let’s connect
Reach out at cheer_dingchiyue@outlook.com if you’d like to walk through the full case study, including the research, design decisions and results behind this project.