ONARVO FRONT DESK CASE STUDY

Cut average first response time from 4 hours to 1.5 hours

Rebuilding Onarvo Front Desk from a basic ticket list into a full-context agent workspace with in-app communication, metrics, AI and automation support for a 30+ person care company.

Quick Summary - If you're in a rush

GOALS

Full context while agents reply
Clear ownership and history
Safe automation managers can control
Keep the system simple for non-technical users

SOLUTIONS

A two column conversation-plus-context layout
Related tickets + team-scoped privacy
Macros + conditional workflows
AI summaries and basic reporting

RESULTS

62% faster first response time
Resolution time down from 3 days to 2
Overdue tickets down from 23% to 9%

The earlier Front Desk

Here's what changed underneath everything else in this case study.

What agents could not do

Single-column ticket details view
No in-ticket conversation or reply
No macros, no workflows. Every action was manual
No reporting at all. No way to see response time, volume, or anything else
No related tickets, no AI summary

REDESIGNING FOR Agent CONTEXT + automation

Two-column layout (conversation + persistent context)
Live conversation with reply, AI summary
Workflows (conditional) + Macros (one-click)
Team-scoped visibility for privacy
Reporting on response time, resolution, and overdue tickets

V1 - Single-column ticket details view

Conversations, and a teardown of what already existed

Requirements came from the PM and directly from care staff, talking through where support was actually breaking down.

Also, I ran an informal teardown of existing tools such as Intercom, Zendesk and Gorgias to understand where the category already was, and where it wasn't.

What none of them had: a way to see a contact's related tickets from inside the one you're looking at. That gap became the related-tickets card.

Inconsistencies across all three: AI summarization was rare, and macro or workflow support varied, some had one, some had the others.

Audits were uncommon: No clear audit trail of who changed what and when

What actually changed, and what it cost

Context cards & related tickets

Agents live on this screen. Conversation stays central; customer, ownership, and related tickets stay visible.

Options Considered

OPTION 1

Move details to a separate tab → loses real-time context

OPTION 2

Decision: 60/40 split. Related tickets only surface within the agent's team.

Tradeoff: Solving for context created a privacy problem the original design didn't have. Teams fixed it.

Trust the summary, not the source

AI summaries help when the thread is long. They become dangerous if agents can't tell AI output from real messages.

Options Considered

OPTION 1

Insert AI summary as a normal message → fastest to scan, easy to confuse

OPTION 2

Decision: Color-coded by source, plus a label - purple for the customer, none for the team, black with an icon.

Tradeoff: Adds a half-second of reading. Removes the risk of treating AI interpretation as the actual record.

Two kinds of automation, on purpose

Workflows = recurring + conditional (match all / match any). Macros = one-click, no conditions.

Decision: Keep them as two separate concepts instead of one flexible builder.

Tradeoff: Two simpler tools felt safer for a non-technical team than one powerful system that could feel technical.

What changed

1.5h

Avg first response (from 4h)

Avg resolution (from 3d)

90%

Tickets with resolution note

What I would do differently

Automate ticket creation

Most tickets still required a manual form. An email-to-ticket path (and later phone/WhatsApp notes) would have removed a repetitive step and reduced the chance of issues being logged late or forgotten.

Revisit the two-concept model

Keeping Workflows and Macros separate matched how managers thought at the time. With more maturity I would test whether a single, well-designed automation builder could cover both without feeling technical.