What 8 Years as a Correctional Officer Taught Me About Customer Success

The Brief

I spent 8 years as a Correctional Officer before I ever stepped foot in a SaaS company. People think it’s a wild pivot, but the truth is, human behavior scales. In a high-risk environment, you learn very quickly that you don’t wait for the incident report to tell you something is wrong. You read the room. You watch the behavior. You catch the pattern early, or you deal with the fallout later. SaaS isn’t any different. Corporate culture isn’t any different. The signals are always there. The hesitation before adoption drops. The silence before churn hits. The passive-aggressive Slack message before the HR escalation. We’ve been trained to manage by metrics, but metrics are just the exhaust fumes of behavior. If you want to predict outcomes, you have to learn how to read the behavioral drift.

The Signal

In a correctional facility, grievance clustering is a reliable precursor to a unit incident. When grievances start clustering around specific areas or times, something is building. In SaaS, I saw the exact same pattern. A highly engaged customer team suddenly went quiet. No feature requests, no support tickets, no emails. The account manager thought they were happy. Two months later, they churned. The silence wasn’t satisfaction; it was the signal.

The System

You have to measure the silence. Establish a behavioral baseline for your accounts. What does “normal” communication look like? When an account deviates from that baseline—whether it’s a spike in aggressive tickets or a sudden drop in engagement—your system needs to flag it as a drift. You don’t need a dashboard; you need an operational radar that alerts you when the room feels off.

The Tactic

Look at your accounts renewing in the next 90 days. Find the ones that haven’t submitted a support ticket or replied to an email in the last 30 days. Don’t send an automated check-in. Send a direct, personal message asking a highly specific question about their workflow. Test the silence.