Escalations Disappeared Into Slack
When an issue needed developer input, agents raised it in a Slack thread. Nothing structurally connected that thread to the Intercom conversation or to the Linear issue that followed. Status depended on someone remembering to report back, and a customer waiting on a fix was invisible unless an agent went looking for them.

No Ticket Architecture Behind the Conversation
Every request lived as a conversation, whether it was a two-minute answer or a bug awaiting a release. There was no distinct state to show that something had left support’s hands, so the inbox gave no honest picture of what was genuinely in progress versus what was waiting on someone else.

Conversations Claimed Manually, Load Distributed Unevenly
Agents picked work from a shared inbox. Distribution depended on who looked first, there was no ceiling on how much any one agent could hold, and the easier conversations were naturally claimed faster than the difficult ones.

Blocked Conversations Consuming Agent Capacity
Conversations waiting on a customer reply stayed open indefinitely. They occupied attention that could not be spent on them, and would have blocked new work from being assigned once automatic distribution was introduced.

Plan Limitations
The workspace was on Intercom’s Essential plan, which excludes Workflows, workload management, SLAs, and the ticketing capabilities that everything above depends on.

Two-Tier Ticket System with Escalation States
Designed a ticket architecture separating customer-facing support tickets from internal escalations, with custom states reflecting where an issue actually sits — Submitted when it needs troubleshooting, Escalated once development has picked it up. State is driven by what happens downstream rather than by someone remembering to update it.

Intercom to Linear Integration
Built a two-way sync using Make. Creating a Linear issue automatically moves the Intercom ticket from Submitted to Escalated and links the issue back to the originating conversation. When the issue is resolved, the outcome posts into the Slack side-conversation thread the agent is already working in — no separate place to check, and no status request to send.

Slack Side Conversations for Developer Escalation
Configured side conversations so escalations reach developers in Slack, where they already work, while staying attached to the Intercom conversation. Support keeps the thread and the customer context; developers are not asked to learn a support tool.

Balanced Assignment with Workload Management
Enabled Balanced assignment — conversations route automatically to whoever currently holds the fewest open conversations — with a cap on concurrent open conversations per agent. Manual claiming was removed, and with it the tendency for easy conversations to be picked first.

Snooze-Based Capacity Recovery
Configured snooze so conversations waiting on a customer leave the agent’s open count and free capacity for new assignment. A workflow monitors them: an automated follow-up after one hour, an automated close with tagging after three, and an immediate reopen the moment the customer replies.

Contact and Conversation Enrichment
Configured person and conversation custom data attributes so context populates automatically when a conversation opens, and key conversation fields are set at creation rather than filled in by hand mid-conversation.

Custom Shopify Review App
Built a canvas app surfacing Shopify review data directly inside the Intercom conversation, giving agents merchant context without leaving the inbox.

Agent Documentation and Rollout
Produced agent-facing documentation covering the new assignment model, capacity limits, when to snooze versus escalate, and the notification settings the change depends on — because configuration nobody understands gets worked around rather than used.