Detect HubSpot Deal Stall Risk & Send Slack Alerts (n8n)

Detect HubSpot Deal Stall Risk — and automatically alert your team in Slack (n8n)

This n8n workflow monitors your HubSpot deals for stall risk, generates a likely diagnosis with OpenRouter (Anthropic models), verifies it with guardrails, and then posts safe, non-duplicate Slack alerts and writes a note back to your CRM.

What this workflow does

• Runs weekly or on webhook: It fetches open deals from HubSpot on a schedule, or triggers when a deal stage change webhook is received (returns 202 Accepted immediately).
• Scores stall risk with deterministic rules: Each deal is normalized into a common schema and scored using weighted factors like activity gaps, time in stage, stakeholder count, close date pushes, and missing next steps.
• Escalates only when it matters: Low/medium risk deals skip LLM analysis. High-risk deals proceed to diagnosis generation.
• Generates and audits a structured stall diagnosis: Using OpenRouter (Anthropic), it produces fields such as likely cause, recommended action, urgency, and confidence, then runs a second call to audit the diagnosis against the raw deal data.
• Applies guardrails before sending: It validates diagnosis fields and performs an Anthropic-powered content safety scan before any Slack message is sent.
• Prevents alert fatigue: Duplicate alerts are suppressed unless the risk score changes significantly or the last alert is older than one week.
• Closes the loop: Sends the verified alert to the deal owner’s Slack channel (or a fallback channel) and writes a CRM note, recording the alert for future suppression logic.

Use cases

• Spot deals that are “stuck” and immediately notify owners in Slack.
• Standardize escalation playbooks for sales ops teams using objective stall-risk scoring.
• Use webhook-triggered updates when a deal stage changes, reducing detection latency.

Technical details

• n8n nodes: if, code, merge, slack, hubspot, webhook
• LLM stack: OpenRouter with Anthropic models (diagnosis + audit) and an additional Anthropic guardrails/content safety check