The two-week notice is a formality. The actual decision to leave happens 60-120 days before the resignation letter. During that window, the departing employee's behavioral patterns shift in ways that are invisible to managers relying on gut instinct but unmistakable in metadata.
Understanding these signals matters far beyond HR. For PE firms and portfolio company boards, unexpected departures of key personnel can derail value creation plans, trigger cascading attrition, and erode institutional knowledge that took years to build. The difference between a planned transition and a surprise departure is often the difference between a minor disruption and a six-figure cost event.
Signal 1: Declining Response Velocity
- The pattern — The employee's average response time to messages and emails increases by 30-50% over a 4-8 week period. Not across the board — selectively. Responses to their direct manager and immediate team slow less than responses to cross-functional requests or company-wide threads.
- Why it happens — An employee who has mentally checked out begins triaging their attention. They maintain minimum viable responsiveness to their immediate circle (to avoid triggering concern) but withdraw energy from broader organizational engagement. The mental calculation shifts from "how can I contribute?" to "what is the minimum I need to do until I leave?"
- The detection window — 60-90 days before resignation. This signal alone has moderate predictive power (it could indicate burnout or a heavy project cycle), but combined with other signals, it is highly reliable.
- What makes it actionable — If detected early, a direct conversation about workload, satisfaction, and career trajectory can sometimes reverse the trajectory. The operative word is "sometimes" — by the time behavioral signals appear, the employee has often progressed past the point where a retention conversation feels authentic rather than reactive.
Signal 2: Shrinking Collaboration Network
- The pattern — The number of unique individuals the employee communicates with regularly drops by 20-40% over a 6-8 week period. They withdraw from cross-functional projects, stop engaging in optional collaborative activities, and narrow their interaction to the people they need for their immediate work.
- Why it happens — Building and maintaining professional relationships requires investment in a future at the company. An employee who does not see a future stops investing. They do not start new cross-team relationships. They let existing ones atrophy. Their collaboration graph contracts toward the minimum necessary connections.
- The detection window — 45-75 days before resignation. Network contraction is one of the earliest and most reliable signals because it reflects an unconscious behavioral shift — people rarely decide to "talk to fewer colleagues." It happens organically as their orientation toward the company changes.
- What makes it actionable — Network contraction is particularly visible and actionable for senior employees and individual contributors who previously had broad collaboration patterns. A staff engineer who used to participate in architecture discussions, code reviews across teams, and mentoring conversations suddenly limiting their interactions to their immediate squad is a high-confidence signal.
Signal 3: Meeting Disengagement
- The pattern — The employee's meeting behavior changes in measurable ways. They decline meetings they previously attended consistently. When they do attend, they contribute less (fewer messages in meeting chats, fewer action items taken, fewer follow-ups initiated). They stop scheduling meetings they used to own.
- Why it happens — Meetings are investments in organizational coordination. An employee who is leaving has no incentive to invest in coordination they will not benefit from. The calculus is unconscious but the behavior is visible: they attend out of obligation rather than engagement, and their participation shifts from active to passive.
- The detection window — 30-60 days before resignation. Meeting disengagement tends to appear later than response velocity changes and network contraction because meetings have a social pressure component — declining a meeting is more visible than being slow to respond to a Slack message.
- What makes it actionable — Meeting disengagement by a team lead or manager is a particularly urgent signal because it cascades. When a manager disengages from cross-functional meetings, their team loses visibility and influence within the organization, creating friction that can trigger attrition among the manager's direct reports.
Signal 4: Knowledge Hoarding or Knowledge Dumping
- The pattern — This signal manifests in one of two opposite ways, depending on the employee's personality and role. Some employees begin hoarding knowledge — answering fewer questions, contributing less to documentation, and keeping work in progress less visible. Others begin a rapid knowledge dump — suddenly documenting everything, writing runbooks they never wrote before, and over-sharing context with colleagues.
- Why it happens — Knowledge hoarders are either consciously protecting their leverage (if they are negotiating internally) or unconsciously withdrawing from the organization's collective knowledge base. Knowledge dumpers are preparing for their departure — they feel an obligation to leave things in good shape, and their sudden burst of documentation reflects a timeline they have already decided on.
- The detection window — 30-60 days before resignation. The hoarding pattern is more common among senior technical staff. The dumping pattern is more common among conscientious managers and team leads.
- What makes it actionable — Both patterns are highly actionable but require different responses. Knowledge hoarding should trigger an immediate assessment of what institutional knowledge the employee holds exclusively and an accelerated knowledge transfer plan. Knowledge dumping should trigger a retention conversation — the employee's behavior suggests they care about the organization's welfare, which means there may be an addressable root cause for their planned departure.
Signal 5: Schedule Pattern Disruption
- The pattern — The employee's working hours shift in ways inconsistent with their historical pattern. Late arrivals or early departures on specific days of the week (interview days). Unusual blocks of "busy" time on the calendar with no meeting details. PTO requests for half-days or individual days rather than vacation blocks.
- Why it happens — Job interviews require time. No matter how discreet an employee tries to be, the scheduling demands of an active job search create detectable anomalies. A person who has worked 9-5 for two years and suddenly has mysterious 11 AM appointments on Tuesdays and Thursdays is not going to the dentist every week.
- The detection window — 30-45 days before resignation. Schedule disruptions typically appear in the final phase of the departure decision, when the employee is actively interviewing rather than passively exploring.
- What makes it actionable — Schedule disruption is the highest-confidence signal because it reflects active, deliberate steps toward departure. By the time this pattern is visible, the employee has typically already decided to leave and is in late-stage interviews. Retention interventions at this stage have a low success rate (under 20%), but transition planning should begin immediately.
The Compound Signal
No single behavioral signal is definitive. Declining response times might indicate burnout. Network contraction might reflect a legitimate shift in project focus. Schedule disruption might be personal. Each signal in isolation has a moderate false-positive rate.
The predictive power is in the combination. An employee showing three or more of these signals simultaneously has an 80%+ probability of resigning within 90 days. The combination eliminates the alternative explanations that make individual signals ambiguous.
What This Means for PE Firms and Boards
Portfolio company boards typically learn about key departures in one of two ways: the CEO mentions it in a board meeting (often weeks after the resignation), or they notice performance metrics declining (months after the departure). Neither provides enough lead time to mitigate the impact.
Behavioral signal monitoring changes this equation. When a key person — the VP of Engineering, the head of Sales, the CFO — begins showing departure signals, the board and operating partner can begin contingency planning: succession identification, knowledge capture, relationship transition, and workload redistribution. The difference between 90 days of preparation and two weeks of notice is the difference between continuity and crisis.
The resignation letter is the end of the story. The behavioral data reveals the plot turns that led to it — early enough to either change the ending or prepare for it.
