Zendesk
Making Usage Visible
TL;DR
I designed a unified usage dashboard inside Zendesk's Admin Center, consolidating data that had been split across two tools. Scoped the MVP with engineering and shipped to 10,000+ customers.
Background
Growing pains
When Zendesk acquired startup Smooch.io in 2019, the platform was rebranded as Sunshine Conversations and became the backbone of Zendesk's messaging infrastructure. As part of the Conversations team, I was brought in to help unify the experience, starting with how customers tracked their usage.
Customer service software
Omni-channel messaging
Powers all messaging services across Zendesk
All conversations captured in one place
Problem
Admins juggled two interfaces for one product
In order to view their complete analytics, customers were required to alternate between two dashboards, creating a fragmented experience.
I was tasked with building a page inside the Admin Center where admins could track their Monthly Active Users (MAUs) and Notifications in one place.
Design constraints
- New dashboard to match Zendesk's Admin Center design system
- No existing data visualization design system to lean on
Process
Mapping, sketching, iterating
Mapping the journey
The admin journey surfaced a key finding: the viewing-usage phase held multiple parallel questions that needed to be answered at once.
Sketching structure
I explored two directions early: a tabbed layout separating MAUs and Notifications, and a continuous scroll showing both together. I kept these rough as the goal was to stress-test the structure with stakeholders before investing in high fidelity.
Iterating the hi-fi
When I presented the first round, senior designers pushed back hard. I took the critique back to Figma, alongside the Admin Center design constraints, and reworked three things:
One page, no switching
Tabs didn't earn their keep when everything could fit on one page. I collapsed them into a single scroll.
Supporting scannability
Metrics weren't scannable enough to be useful. I pulled the current-period numbers into three widgets up top, readable at a glance.
Content boxing
Analyzing the Admin Center's design system, I wrapped sections in matching card conventions so the page felt native.
Scoping to what could ship
Due to its high-priority, my PM flagged that data visuals weren't feasible for initial release. This forced me to figure out what the page truly needed.
Final Product
The MVP that shipped
Distinct widgets for the current period, table underneath for six months of MAU and notification history, all on one page. Inline links explain what MAUs and Notifications are and what counts toward each.
The shipped version, table-first.
Exploration
Visualizing the full picture
Once the MVP was scoped and approved, I was tasked with exploring what a visual version could look like as a future enhancement. I also designed for the full range of states a customer might encounter.
Normal usage
6-month bar graph and data table.
Plan limit reached
A message box prompts users to contact support to adjust their plan. Affected metrics are called out in orange.
Conversations paused
When usage is severely over the limit, conversation services are paused and a prominent error banner drives users to contact support.
Impact
Shipping the small (but mighty) version
I designed the full dashboard vision, then cut it with engineering to what could ship. The table view launched, unblocking customers who had been jumping between two tools. I mapped API and integration schemas to accelerate the handoff, finishing ahead of schedule.
10,000+
Users reached at rollout
15%
Faster feature completion
Reflections
Beyond the design file
Leading a real project taught me that product design lives as much in the conversations and tradeoffs as it does in Figma.
Constraints clarify
Designing within real constraints forced me to get clear on what the page actually needed versus what would be nice to have, and how to make the case for the best experience within those boundaries.
Embracing the loop
Each round of feedback was a chance to ask the right questions, navigate ambiguity, and ultimately pull the design somewhere better.
thanks for reading (✿・‿・)ノ゛
Up next
FigBuild 2026
