The email lands on a Tuesday. Subject line: "My two-week notice." It is not from the employee struggling to meet their KPIs. It's from the team lead. The top engineer. The project manager who held three workstreams together.
That's what regrettable attrition looks like, according to new research from Spring Health: not the departures you planned for, but the ones you were counting on not happening.
The goal of HR isn't to stop all turnover. It's to identify and prevent the loss of the people you most want to keep — the high performers, the institutional knowledge carriers, the culture drivers — whose departures seem preventable in retrospect, because they were. The signals were there. Nobody was reading them.
Why Exit Interviews Have Never Been Enough
For decades, the exit interview has been the primary tool for understanding attrition. HR teams diligently code the reasons: compensation, career growth, management. They report them upward. Leadership nods.
The problem is structural. Exit interviews are lagging indicators. By the time you're running one, the talent has already left. You are analyzing the past rather than influencing the present.
What Spring Health calls a "leading indicator system" is different in kind, not degree. It looks for signals across the full employee lifecycle — onboarding friction, performance review nuances, benefit utilization patterns — that appear long before a resignation letter is drafted. The path to a regrettable exit is almost always a slow, silent slide. The system's job is to see the slide beginning.
The Four Pathways Nobody Tracks
Employees rarely quit on a whim. Spring Health's framework identifies four common pathways that lead to regrettable attrition — each one detectable before it reaches the exit.
Burnout to disengagement to job search. The pathway begins with unmanageable workloads or insufficient recovery time. The employee powers through — until they hit a wall. Motivation drops. Productivity slows. Psychological disconnection sets in. Once disengagement takes hold, the exit becomes the next logical step. Burnout doesn't announce itself; it shows up in declining output quality, rising absence patterns, and a gradual retreat from optional engagement.
Manager friction and the performance spiral. Research cited by Spring Health shows that managers affect employee mental health as much as spouses do. The pathway typically begins with unclear expectations or absent feedback. An employee grows unsure of their standing. A surprise negative performance review damages trust. The employee starts looking for a culture where they feel valued. The manager — often burned out themselves — never saw it coming because they avoided the conversation that could have changed the trajectory.
Unplanned mental health leaves as "retention smoke." Unplanned leaves of absence are often a final distress flare. When a high performer suddenly needs time off for mental health reasons, it signals that the environment has become unsustainable for them. Spring Health's framework treats unplanned leaves not as compliance events but as smoke — evidence of fire somewhere in the organization. The question worth asking isn't just how to manage the leave, but which team, manager, or structural condition produced it.
Financial strain and life instability. Employees bring their whole lives to work. Chronic financial stress — inflation, debt, caregiving costs — creates a cognitive load that makes it difficult to focus. The employee seeks a "safety raft": often a new job with a signing bonus or marginal salary increase, even if they genuinely like their current role. This pathway is frequently invisible to managers until it's over.
What Leading Indicators Actually Look Like
The measurement system Spring Health recommends doesn't require exotic data. The leading indicators are already in the operational data most HR teams collect:
- Sudden productivity drops — not the gradual decline of a performance issue, but a notable shift in output or quality from someone previously reliable.
- Attitude changes and withdrawal from team activities — the team member who used to show up early to meetings, who has stopped. The contributor who went quiet in channels they used to drive.
- Unusual leave patterns — both the absence of PTO usage (a burnout warning) and sudden sporadic absences (a distress signal) tell different parts of the same story.
- Low scores on pulse surveys around stress and workload — not satisfaction scores, which lag further. The specific signal around cognitive load and workload manageability.
No single indicator is diagnostic. The pattern across indicators — especially when correlated with manager tenure, team conditions, and life stage — is where the signal lives.
The Missing Layer in 2026: AI Infrastructure Risk
There is a dimension to regrettable attrition in 2026 that Spring Health's framework doesn't yet fully capture: the AI infrastructure that walks out the door with your best people.
High performers today are not just carrying institutional knowledge. They are carrying active AI infrastructure — the Copilot workflows they built, the automations their team depends on, the prompts refined over months of iteration. When they leave, those don't automatically transfer. The junior employee who relied on their AI-augmented output loses a capability multiplier, not just a colleague.
The organizations that will manage regrettable attrition in 2026 are the ones that know, before the exit email arrives, which employees are both flight risks and AI-critical. That intersection is where retention planning has to live now.
The signals are already in your data. The question is whether your system is reading them — or whether you're still waiting for the two-week notice to tell you what happened.