| name | chief-escalation |
| user-invocable | false |
| description | Synthesize customer escalation data from support tools into actionable intelligence. Use this skill when the user asks about customer escalations, support trends, customer health, at-risk accounts, ticket volume, or the support-to-build pipeline. Triggers on "escalation digest", "support summary", "customer health", "at-risk customers", "what's happening in support", or when preparing board materials or customer reviews. Pulls from Pylon (tickets), Slack (support channels), Notion (customer docs), Jira (engineering issues), and HubSpot (deal/renewal data). |
Customer Escalation Synthesizer
You monitor and synthesize customer support data into CEO-level intelligence. Pylon is the primary data source — its MCP tools are available directly.
Pylon MCP Workflow
Pylon tools are available directly — no ToolSearch needed.
Data Collection Steps
- Get all open issues across all accounts —
mcp__pylon__search_issues with:
states: ["new", "waiting_on_you", "waiting_on_customer", "on_hold"]
limit: 100
- Page through results using the returned
cursor if needed
- Get recently closed issues (for resolution wins and trend analysis) —
mcp__pylon__search_issues with:
states: ["closed"]
created_after: 7 days ago (RFC3339 format, e.g., "2026-04-03T00:00:00Z")
- Get issue details for escalations — For any issue that needs CEO-level context:
mcp__pylon__get_issue for full details (body, custom fields, account info)
mcp__pylon__get_issue_messages for the full conversation thread
- Search accounts for context —
mcp__pylon__search_accounts to:
- Identify accounts with the highest issue volume
- Get account owner, tags, and custom fields
- Use
mcp__pylon__get_account for full details on flagged accounts
- Cross-reference with HubSpot — For each flagged account, search HubSpot for:
- Deal size (ARR/ACV), renewal date, expansion opportunities
- This is what turns a support ticket into a CEO-level escalation
Filtering for CEO Attention
Use Pylon issue states to triage:
waiting_on_you issues aging > 48 hours = response SLA risk
- Multiple
new issues from one account in a week = systemic problem
on_hold issues aging > 7 days = stale escalation
- High issue volume on an account with an upcoming HubSpot renewal = at-risk
Week-over-Week Comparison
To calculate trends:
- Run
mcp__pylon__search_issues with created_after/created_before for the current week
- Run again for the prior week
- Compare: total volume, volume by account, volume by tag/category, state distribution
Capabilities
Weekly Escalation Digest
Pull from Pylon (primary) and cross-reference with HubSpot, Slack, Jira. Produce:
- Top issues by severity, customer, and category
- New escalations since last digest
- Aging escalations (unresolved > 7 days)
- Resolution wins (what got fixed)
- Trend vs. prior week: volume up/down, category shifts, emerging patterns
Ticket-to-Build Pipeline
Identify support patterns that should become product:
- Issues that appear across multiple customers
- High-volume ticket categories that a plugin or feature could eliminate
- Quantify: "X tickets/week across Y customers" for prioritization
- Flag opportunities where CS/support could build the solution directly rather than triaging repeatedly
- Cross-reference with Jira: search Notion (Jira connected source) for existing engineering tickets that match the pattern. Avoids recommending work that engineering already has in flight.
Customer Health Scoring
Cross-reference support data with business data:
- High ticket volume + upcoming renewal = at-risk flag
- High ticket volume + new customer = onboarding issue flag
- Low ticket volume + high engagement = healthy
- Search HubSpot for renewal dates, deal size, expansion opportunity
- Expansion-at-risk detection: If an at-risk account is also flagged as an expansion opportunity in HubSpot, escalate with a different playbook — this needs a save-and-grow strategy, not just a retention play. Note the expansion ARR at stake alongside the base ARR.
At-Risk Customer Alert
Flag customers where:
- Support volume spiked >50% week-over-week
- Severity 1 tickets are open > 48 hours
- Multiple open tickets across different categories (systemic issue)
- Renewal is within 90 days AND support volume is elevated
- Expansion opportunity in HubSpot AND support volume is elevated (flag as "expansion at risk")
Output
Use /chief-memo conventions. Structure:
- Summary — 3-4 sentences: overall support health, biggest concern, biggest win
- Key Metrics — Open issues (by state), closed this week, WoW trend, top accounts by volume
- Escalations Requiring CEO Attention — Only the ones that matter at this level. Include Pylon issue number, account, state, age, and the HubSpot deal size + renewal date.
- At-Risk Accounts — Table: account name, open issues, "waiting_on_you" count, deal ARR, renewal date, risk factor
- Build Opportunities — Support patterns that should become product (tag/category clusters across accounts)
- Resolved This Week — Wins worth knowing about
Jira Cross-Reference (Engineering Source of Truth)
When surfacing ticket-to-build opportunities or escalations with a technical root cause, cross-reference against Jira via Notion's connected source:
notion-search: query="jira [issue keyword or component]", query_type="internal"
Use this to:
- Check if an engineering ticket already exists for the reported problem (avoids creating duplicate work)
- Find the assigned engineer and estimated fix timing
- Determine if an open Jira issue is causing the spike (and escalate with the ticket number)
- For P1-equivalent escalations, note the Jira issue status in the "Escalations Requiring CEO Attention" section
Engineering data source: Jira (via Notion AI) — not a direct Jira MCP. Query through Notion with query_type="internal".
MCP Dependencies
| MCP Server | Role | Required? |
|---|
| Pylon | Primary source for all support tickets, account data, issue states | Required — skill cannot function without it |
| HubSpot | Deal size, ARR, renewal dates for at-risk context | Required — needed to turn tickets into CEO-level escalations |
| Notion | Jira integration (engineering issues), customer docs, handbook context | Required for ticket-to-build and technical cross-reference |
| Slack | Support channel discussions, internal escalation threads | Optional — fallback if Pylon is unavailable; enriches context |
Rules
- Only escalate to the CEO what genuinely needs CEO attention. Filter aggressively.
- Always include the customer name, deal size (from HubSpot), and renewal date for context
- If Pylon MCP is unavailable, search Slack support channels as a fallback. But Pylon is the primary source — always try it first.
- Quantify everything — "several customers are frustrated" is useless. "4 customers filed 12 tickets about lab ordering in the last 7 days" is actionable.