Cutting complaint resolution time by 32% for support agents handling 34+ live tickets at once
DURATION
2022–2023
ROLE
Sole Designer
COLLABORATORS
Support team, Product lead & 5 Engineers
PLATFORM
Web
At Avhan, I designed Jodo — the support workspace our agents live in for 40+ hours a week. Agents were resolving billing disputes, network outages, and KYC issues across voice, chat, email, and SMS, while racing an SLA clock on every ticket. My goal was to use everything I know about visual hierarchy and information density to build one surface that holds the queue, the live conversation, the customer's full history, and the next best action — without an agent ever switching tabs.
The outcome was a measurable shift in how fast the team could work. We reduced average time to resolution by 32%, kept SLA breaches down even as ticket volume climbed, and gave agents a single source of truth for every customer, trade, and conversation on the platform.
One queue for every channel, sorted by what's about to breach
Resolve a dispute without leaving the conversation
Every customer's full history, surfaced before they finish the sentence
Why build one workspace instead of stitching tools together?
A single agent surface would hit three targets with one arrow — collapsing the tool-switching that slowed every ticket, getting agents to a resolution faster, and freeing engineering from one-off data lookups.
A unified surface for agents
The workspace became the one place agents worked, replacing the constant jump between dialer, billing, CRM, and knowledge base to assemble a single answer.
Faster, better-informed resolutions
With history, sentiment, and policy in one view, agents handled disputes in one pass — and customers felt the difference in how prepared every agent was.
Less load on engineering
Surfacing transaction logs and account context directly in the tool removed the steady stream of "can you check the backend?" requests, letting engineers stay on the product.
Sitting on the floor with the people who'd use it
I spent time alongside the support agents during live shifts, watching how a single ticket actually got resolved end to end. Following a complaint through the queue, into the call, across the tools they pulled open, and out to a resolution showed me exactly where the time leaked — and which moments mattered most to get right.
What I observed,
Beyond manually tracking where each ticket stood, agents moved between separate tools to pull billing, transaction, and conversation data — re-reading context the system already had, on every single ticket.
What I was constrained by,
To keep front-end engineering light, I had to repurpose and adapt existing components into the workspace rather than build net-new, and hand off in a form the team could ship quickly.
Crafting interactions and components for a workspace agents could move fast in
Even though this was an internal tool, I was committed to a workspace that maximizes productivity and minimizes development complexity. That starts with intuitive, efficient components.
A status system that lets agents triage the whole queue at a glance.
One tab switch to scope the queue by voice, chat, email, or SMS.
The context panel populates the moment a ticket opens — no lookup needed.
Policy-matched next steps the agent applies with one tap, never automatically.
A running read on tone, flagging the moment a call shifts.
Steps check off as the call progresses, so nothing gets missed under pressure.
Transaction IDs and key terms highlight live in the transcript.
The wrap-up drafts itself from the transcript, editable before submit.
Supervisors ranked by best match — load, AHT, CSAT, and accept time.
A supervisor's real-time read on every agent, queue depth, and breach risk.
Raising the bar on the support experience
The workspace lifted the team's day-to-day speed and held service quality steady even as volume grew.
-32%
in average time to resolution
[X] min
average first response time
[X,XXX]
tickets resolved in the busiest month
"[Quote placeholder — support lead]"
[Name] · [Role], Support
"[Quote placeholder — operations]"
[Name] · [Role], Operations
Looking to add a designer who builds tools the team actually wants to use? Let's talk.
Resume
[email]