Unifying customer feedback across 19 products and 9 channels

A fully configurable internal tool that brings together customer feedback from 19 products and 9 channels, making it easy for teams to explore, analyze, and act on what users are really saying.

Client
Atlassian
Sector
Enterprise SaaS
Role
UX UI, User Research
Focus
Internal tools, Data visualization, Information architecture
Year
2023

Turning scattered feedback into one actionable view

Atlassian needed a way to turn scattered user feedback into a centralized, actionable view.

We partnered with their team to design an internal tool that connects the dots, giving product managers and account managers the ability to explore raw feedback, identify trends, and measure impact across all products and channels. The hard part was never collecting feedback, it was making sense of it once it arrived from everywhere at once.

Through iterative research, wireframes, data visualization, and multiple rounds of usability testing, we built a flexible, user-centered solution grounded in Atlassian's design system. From customer sentiment to feedback themes, the final product empowers teams to prioritize with confidence and finally close the loop.

Fig 01 The Customer Feedback Ecosystem dashboard. Sentiment and volume over time sit alongside emergent themes and the KPIs that tell PMs where to look first.
70%
Faster workflow
Time to identify feedback trends
50%
Fewer clicks
To trace feedback to its source
+30%
Satisfaction
From product managers using the tool
+40%
Confidence
In data-driven decision making

A chart that holds more than one dimension at a time

This type of chart is multi-dimensional, versatile, and allowed more data points than other charts. Since Atlassian decided to build their own tool rather than buy one, we had more creative license than usual to break the traditional data visualization boundaries.

Tree maps are used throughout the design to communicate feedback volume and negative sentiment change through either themes, customers, channels, or products. Size encodes the volume of feedback; color encodes the negative sentiment change, so a single block tells you both how loud a topic is and how fast it is turning sour.

Fig 02 The tree map encoding. Block size shows the volume of feedback, color shows the negative sentiment change, and the annotations connect each block back to its theme, channel, and the products it touches.

Understanding the why behind feedback

One of the goals of the project was to help product managers move from a high-level overview of feedback trends to the raw details of individual comments. The view reveals patterns, tracks sentiment, and connects insights directly to users and products.

From the main view, product managers can:

  • See how many tickets each theme has created
  • Identify key phrases driving those themes
  • Track sentiment and feedback volume over time
  • Assess how recent updates have influenced sentiment
  • Understand which customers and products are most affected
  • See which channels users use to share feedback

When more context is needed, they can drill down into verbatim feedback, view user profiles, and even reply or schedule time with users to learn more about their experiences.

Fig 03 The Themes view rolls feedback up into trends: tickets generated, net sentiment, and volume by channel and product, so a PM can see a theme's trajectory before reading a single comment.
Fig 04 The Raw Feedback view drills all the way down to the verbatim comment, filtered by theme and phrase, where a PM can reply or book time with the user without leaving the tool.

Information architecture

The feedback needed to be divided into several slices at a high level first, and then a more detailed view. The deepest level was Raw Feedback, which was the most interesting to most of the users, since it provides an unbiased, unfiltered view into the verbatim feedback shared by users. It also showed how that feedback was rolled up into the other slices, so a user could always trace a trend back to the comments underneath it.

Fig 05 The information architecture. The Dashboard fans out into five slices, Themes, Channels, Products, Customers, and Raw Feedback, each drilling down to a single detail view, with Saved Views and Preferences kept to the side.

Design approach

We opted for flat, minimal styles for any new styles that were not already in the Atlassian design system, for example, the data visualizations. We also opted to avoid the traditional red, yellow, green stoplight pattern to show positive and negative trends, in order to be accessible and color-blind friendly.

Fig 06 The same dashboard, viewed normally and through a deuteranomaly simulation. Avoiding the stoplight palette keeps positive and negative trends legible for color-blind users.
Accessibility is not a finishing pass. It decided the palette before the first chart was drawn.

Design artifacts

In addition to final product design, we provided user flows, information architecture, wireframes, prioritized usability testing findings and recommendations, and design system updates. The deliverable was a system the Atlassian team could keep building on, not a single set of screens.

Color palette

#091E42
#403294
#0747A6
#008DA6
#006644
#FF8B00

Read another case study

More selected work across enterprise AI, narrative intelligence, and complex internal systems.