Feedback From Customer Support During Incidents

Support sees the customer-facing reality. The feedback loop that helps the IC.

Channel

Support sees customer-facing reality before engineering does. Customers tell support what they actually experienced; engineering only sees what the alerts caught. The channel between the two needs to be direct (not routed through ticketing) and named (a specific liaison the IC can address).

• Direct support-to-IC channel. Dedicated incident Slack channel between support and IC. Not through ticketing, not through the regular escalation path.
• Real-time signal. Live customer reports as they arrive. Customers tell support things engineering has not seen yet.
• Named support liaison. One responsible support contact per incident. Catches diffuse signal where nobody knows who is the source.
• IC acknowledges every update. Visible "got it" from the IC keeps the support side engaged. Silence on the engineering side burns the signal channel.

Filter

Support filters before forwarding. Anomalies and patterns reach the IC; raw individual reports stay in the support queue. The IC has to trust the filter or the channel becomes noise; support has to take the gatekeeper role seriously.

• Support filters first. Anomaly-or-pattern threshold for what reaches the IC. Not every individual report.
• IC trusts the filter. No second-guessing the support team's judgement during the incident. Forces support to take the gatekeeper role seriously.
• Concrete cluster threshold. "5+ similar reports in 30 minutes" or equivalent. Catches inconsistent escalation across incidents.
• Structured update format. Count, customer segment, geography per update. Supports IC pattern recognition over the wire.

Act

Support signal updates the IC's mental model and sometimes shifts the investigation entirely. The customer-facing reality occasionally catches the real cause before the engineering instrumentation does, especially for issues that show up in user behaviour before they show up in metrics.

• Updates the IC's mental model. Each support update is new evidence that the IC integrates into the working hypothesis.
• Shifts the investigation when needed. "We were chasing the wrong thing" reset based on support signal. Customer reality wins.
• Visible IC response per update. Acknowledgement plus action. Keeps the feedback loop closed.
• Support-feedback section in the postmortem. Recap of support's contribution per incident. Supports cross-team learning and reinforces the channel.