Your Last 90 Days at a Company Start Before You Resign: Understanding What You Want, Handing Over Responsibility, and Leaving Well

On the company calendar, the final chapter begins when someone says, “I’m leaving.” In a career, it starts earlier, when routine frustration turns into a question. What exactly is no longer working: the work, the role, the manager, the company’s stage, my own exhaustion, or the direction in which I no longer want to grow? Until that question is unpacked, resignation may end an unpleasant condition without moving the person toward a better future.

It is more useful to treat departure as a five-part transition: diagnosis, options, decision, transfer, and renewal. The stages overlap. A conversation about an internal role can run alongside market research. A responsibility map is useful before formal notice. External interviews may reveal that the core mismatch is not the employer but the kind of work the person keeps selecting.

This frame returns agency to the person making the decision. Leaving stops being a reaction to the latest conflict, a delayed promotion, or one brutal quarter. The company also gets a chance to do something more useful than retain someone at any price: it can test whether reciprocal value can be redesigned. If it cannot, both sides still have time to move the work without heroics or a search for villains.

The Book Cube review of Designing Your Work Life presents working life as something to design: understand the situation, generate options, prototype them cheaply, and only then implement a choice. That discipline matters especially when leaving. You do not need one perfect answer at the beginning. You do need to avoid replacing investigation with the fastest way to stop feeling uncomfortable.

Understand what is failing → describe the desired work → test internal and external options → decide → transfer responsibility → begin the next chapter

Ninety days is a useful planning horizon, not an instruction to conceal a decision for three months or impose an arbitrary notice period. 010101When circumstances allow less time, the sequence still helps but each stage contracts. When there is more time, succession can be built before any decision to leave—as ordinary healthy leadership practice. The point is to avoid compressing the entire transition into the final overloaded days, when every unresolved commitment must either be abandoned or carried into the next chapter.

“I cannot keep doing this” describes the intensity of a signal, not its cause. It may contain several incompatible problems. Someone lacks autonomy even though the mission and colleagues still matter. A role demands constant outward communication while energy comes from deep technical work. The company changed strategy and the old source of meaning disappeared. Workload surged temporarily after a crisis. Or the environment repeatedly violates the person’s values. Each account points to a different intervention.

Begin with observations rather than a verdict. For several weeks, note which events create or drain energy, where curiosity appears, where avoidance begins, which conversations clarify the work, and which create helplessness. The energy log itself comes from Bill Burnett and Dave Evans; Will Larson separately advises executives not to decide from one bad week but to examine a sustained pattern. A log does not decide for you. It separates a repeating mechanism from a vivid episode.

Then interrogate those causes with four diagnostic questions. The first is energy and load: is the exhaustion caused by a temporary peak or the normal shape of the job? 020202Stulberg and Magness show in Peak Performance that load without recovery burns out even people doing work they love. That is a useful correction to the diagnosis: sometimes the role is fine and the regime is broken. I reviewed the book on the channel.Book Cube · Peak Performance The second is work design: could different tasks, relationships, or ways of seeing the outcome change the experience? Amy Wrzesniewski and Jane Dutton’s work on job crafting describes changes to those task, relational, and cognitive boundaries. The third is autonomy, competence, and relatedness. Self-Determination Theory identifies them as basic psychological needs; here they serve as questions, not a clinical or resignation test. The fourth is trajectory: what are you learning, and does it move you toward the person you want to become?

• —Which parts of the work reliably create and drain energy?
• —Which activities should grow, and which should stop defining the role?
• —Where is the missing ingredient: decision authority, mastery, or meaningful connection?
• —Which capability does this environment develop—and which does it now constrain?
• —What observable change would make staying a reasonable decision?

Do not turn the diagnosis into an indictment. “My manager is bad” is almost impossible to test or use. “Priorities changed three times without a discussion of consequences while my role retained accountability for the dates” is an observation that supports a conversation. “The company does not develop me” is too broad. “The coming year contains no opportunity to lead leaders or own a commercial outcome, and those are the two capabilities my path now requires” is actionable.

A useful diagnosis ends not with the word “resign” but with a minimal, testable design problem. Restore prioritization authority. Change the share of operational work. Gain experience at another product stage. Stop representing a strategy you can no longer support honestly. Only then can you ask whether the problem requires a new employer or can be solved where you already are.

The next mistake is translating dissatisfaction too quickly into a job title. “I want to be a CTO,” “I want to return to engineering,” or “I want a larger company” sounds directional while hiding the actual work. One CTO builds an organization of two hundred engineers, another sells a technical vision to customers, and a third remains the chief architect. Returning to engineering may mean individual contribution, technical leadership without direct reports, or simply a season away from people management.

First collect evidence from your own path. Which decisions did you make particularly well? When did you receive the most useful feedback? Which results would you gladly produce again, and which would you never again buy with the same amount of attention? What relationships improved your work? In Managing Oneself, Peter Drucker asks people to know their strengths, values, working style, and place of greatest contribution. This is not a search for an immutable identity. It is design material for the next chapter.

Check whether you are trying to solve a fixed constraint. Designing Your Work Life calls these “gravity problems”: they are real, but the current framing offers no action. You cannot make a mature company become an early startup again, reverse a completed acquisition, or obtain an executive remit inside a structure where the function does not exist. You can acknowledge the constraint, change the question, and choose a different context.

Describe the desired work across several dimensions: problem, scale, share of technical and organizational work, company stage, decisions owned, relationship with the manager, feedback speed, acceptable uncertainty, way of life, and capability to develop. Add anti-goals—activities you are not willing to make central again, even for a larger title. The result is not a fantasy job description. It is an evaluation map.

Build two or three future versions. One continues the current path at greater scale. A second changes the form of influence—for example, from leading a large function to technical strategy or product. A third explores a more ambitious turn. For each, ask which experience already supports it, what evidence is missing, and which small experiment would produce that evidence. An informational conversation, a short contribution to a neighboring initiative, or shadowing another function is cheaper than accepting a job simply to test a fantasy.

This exercise also protects against treating a raise or title as sufficient evidence of growth. 030303Orosz makes a simple point I keep repeating to mentees: the engineer owns the career, not the employer. Hence the habit of keeping your own record of outcomes — it serves the internal conversation and the external interview equally.Book Cube · The Software Engineer’s Guidebook Money matters. A long-horizon choice also includes meaning, relationships, health, mastery, freedom, and the future usefulness of the experience. “What can the company do for me?” becomes a serious question only when you can name the next capability you need and the conditions under which it can grow.

Once the desired work is visible, the external market feels like the obvious next stop. Look inside first. The company already knows your history, and you understand its product, people, constraints, and informal relationships. That shared context lowers the cost of orientation and may allow substantive work to begin earlier. The organization preserves knowledge and trust; the employee preserves part of their accumulated influence and can test a new kind of role without replacing the entire environment.

There are at least four internal routes. Renew the purpose of the existing job. Redesign its tasks and boundaries. Move to another team, product, or geography. Or change professional tracks while remaining in the same organization. The Designing Your Work Life review names these routes Re-enlist, Remodel, Relocate, and Reinvent. The labels matter less than escaping the false binary of endurance or resignation.

Search for the intersection of three maps, not merely an open requisition: an outcome the company needs, a capability you want to develop, and a sponsor able to grant the mandate. Sometimes the desired position already exists. Sometimes it can be formed around an important problem that currently lacks an owner. A custom role invented only to retain someone, without a customer, outcome, or decision rights, usually recreates the old dissatisfaction under a new label.

Frame the conversation as a hypothesis to test rather than an ultimatum. Present the diagnosis without performing an irrevocable decision: what no longer works in the current design, which outcome you could produce in a different configuration, what a bounded experiment might show, and which authority it requires. Agree on a next step—a meeting with a sponsor, temporary work on an initiative, a concrete role proposal, or an explicit answer that no such opportunity exists now.

Matthew Bidwell compared external hires and internal movers at one U.S. investment bank. In that specific setting, internal candidates held advantages in early performance and retention, while external hires began with higher pay. The study is useful evidence that local context has value, but it is not a universal law. One firm, one industry, and important differences among types of internal moves do not justify saying that internal candidates always win.

Apply the same rigor to an internal offer that you would to an external one. Who evaluates the outcome? Which former duties disappear? What happens to the team? Who protects the remit when priorities conflict? When will the agreement be reviewed? 040404A counteroffer made only after notice is not automatically bad. Evaluate it as a new role: did the mechanism causing the problem, the mandate, and the resources change—or only the price of the old work? An internal move is valuable when the design of the work changes, not when it merely postpones the same decision.

Exploring options should not become indefinite waiting for perfect certainty. Set a decision date and evidence criteria. What must become true for staying to make sense? What counts as a successful internal experiment? How long are you willing to wait for the remit to be confirmed? Will Larson suggests choosing a future decision point when constant reconsideration has itself become a source of exhaustion.

There are three honest outcomes. Stay in the current role because the diagnosis revealed a temporary problem or because your own perspective changed. Move internally under explicit conditions. Or prepare to leave. The unhealthy outcome is choosing one formally while continuing to inhabit another: accepting an internal move while privately treating it as a brief stop; announcing a departure while trying to control the team’s future; or staying while reopening the decision every morning.

If you stay or move, renew the reciprocal contract. This is not a legal document. It is an explicit understanding of mutual obligations. Denise Rousseau describes a psychological contract as an individual’s belief about exchanged promises between person and organization. Invisible promises are often the ones that fail: “I will be allowed to build the function,” “she will keep personally containing crises,” or “the budget will appear in six months.” Turn them into testable agreements.

Two symmetric questions

Most career conversations emphasize “What can you do for the company?” The question is necessary: a role exists to produce an outcome, not solely to develop the person holding it. Without the other half, however, the contract is one-sided. What can the company do for you? Not as an unlimited collection of benefits, but as access to problems, scale, feedback, and relationships aligned with a long-term development plan.

• —Which outcome does the company receive from my next chapter?
• —Which capability, reputation, or area of mastery do I gain?
• —Which decisions and resources make the exchange credible?
• —What do the two sides stop expecting from each other?
• —When and with which evidence will we revisit the agreement?

If the answer is departure, the reciprocal contract changes too. Before notice, you still own the role and should decide in the company’s interest. After notice, the center of gravity moves toward continuity: help the future owner understand the system, avoid creating new long-lived dependencies, and do not use the departure as one final chance to force through reforms. This is not an abandonment of responsibility. It is a different form of it.

Internal exploration and the external market do not need to occur in a strict sequence. Conversations with other companies test whether the desired role exists beyond your imagination, how the market names the work, and which gaps it sees in your experience. An internal experiment tests whether a different employer is needed at all. Parallel work is useful as long as it does not become a double game or violate obligations to the current company.

Do not search for “the same thing without today’s pain.” Turn the desired-role map into research questions. Why is the position open? What does an ordinary week look like? Which decisions belong to the role? What did the previous person learn, and why did they leave? Which part of the work is unusually difficult at this business stage? How will the company invest in the leader’s growth after hiring? 050505I spent years building the hiring funnel for technical leaders and presented it at Teamlead Conf. Knowing what each stage actually checks helps more than the hiring side: a candidate who understands it asks far sharper questions.Book Cube · hiring technical leaders The answers should show not only whether the offer is attractive, but how it connects to your trajectory.

The companion longread on the CTO’s first 90 days begins before day one: with market research, testing the real problem, interviewing in both directions, and negotiating the mandate. Those actions are simultaneously the last steps of the old transition and the first steps of the new one. The stage deck offers a shorter version of the same model.

Preserve clean professional boundaries while still employed. Do not use confidential material in a portfolio, move internal data into personal systems, or present a team’s work as an individual achievement. Cases can explain the context, your role, the decision, an observable result, and the lesson while removing sensitive detail. Questions involving contractual restrictions, intellectual property, notice, or overlapping employment depend on the agreement and applicable law.

Searching while employed sometimes creates negotiating room because the person need not accept the first escape from discomfort. At other times, depletion is severe enough that rest matters more than leverage. Larson names this fork for executives. There is no universal answer. Ask which condition will improve the quality of the next decision: a parallel search now or deliberate space between chapters?

Search for the next development context, not merely the next employer: the company’s problem + your contribution + the capability you want to build

Responsibility transfer starts not with a documentation list but with a map of what the role actually holds. A job description names broad areas while hiding commitments, decisions, and relationships. Who promised a customer an outcome? Which risks are raised to the executive team every month? Who resolves a product versus reliability conflict? Which operating review works only because you remember its unwritten sequence?

Divide the map into six layers: role outcomes and boundaries; active initiatives; recurring cadences; stakeholders; material risks; knowledge and access. For each item, name the future owner, backup, next action, and acceptance check. When the owner is unknown, that is not a documentation gap. It is a management decision that must be escalated.

Do not promise to finish everything. That usually creates false heroics: the departing person clears small tasks, delays difficult transfers, and leaves a clean personal list but a fragile system. Sort work into four groups: complete before departure; transfer with a clear next action; stop deliberately; replan with the new owner. For unfinished work, naming the decision and its context is more valuable than creating the appearance of completion.

Align the transition with the manager in writing: dates, communication order, future owners, critical risks, decisions that require executive participation, and a review cadence. A short note protects both sides from competing memories. It also reveals where the company must act. A departing leader can prepare succession options, but appointing the successor and assigning authority belongs to the organization.

For a leader, succession should not begin with resignation at all. Larson recommends developing reports toward pieces of the role, exposing possible successors to senior leadership, delegating recurring meetings, and using extended absence as a resilience test. This does not make the leader less valuable. It releases attention for the next scale of work and lowers the cost of any eventual transition.

The test of a useful plan

A transition plan is good not when it contains many rows, but when it supports decisions under pressure. Any participant should be able to see what may stop, who now owns the question, which decision comes next, where the source of truth lives, and when to escalate. If the answer is still “ask the departing person,” the transfer has only been described.

A document transfers information, not the ability to act. Ambiguous decisions are particularly hard to move: a conversation with a critical customer, assessment of a leader, response to an incident, or tradeoff between speed and resilience. The unit of transfer should therefore be a cycle rather than a file: explain the context, work together, let the new owner perform the task, discuss the decision, and explicitly accept the result.

Build a compact canvas for each critical area. What outcome defines success, and what is outside the boundary? What is the normal cadence? Which decisions recur? Where are the facts? Which people influence the outcome, and what have they been promised? Which early signals indicate danger? What has failed before? What may the new owner change immediately, and what requires joint commitment?

The practical protocol is straightforward. Inventory the work and prioritize areas by the cost of interruption. Name an owner and confirm their authority. Run one cycle together, explaining not just the action but the reasoning. Let the new owner run the next cycle while you observe and answer questions. At the end, they explain the boundaries, risks, and next steps in their own words, and the responsible manager confirms the transfer.

Cornell’s knowledge-transfer guidance emphasizes continuity rather than an attempt to document everything: responsibilities, recurring work, active projects, workflows, systems, materials, partners, and risks. That principle has a practical consequence: write for a specific next action. Ten pages of service history may be less valuable than one page explaining what to inspect weekly, which signal means danger, and who makes the decision.

Transfer access through the organization’s normal process, never by sharing credentials. Verify that the right people can actually open dashboards, documents, and operating systems. Assign owners to shared materials. The company should remove or retain access after departure under its own policy. This looks like an administrative detail, yet a hidden dependency on a personal account can turn a careful transition into emergency recovery.

Knowledge moves through conversation, and that is a skill of its own. 060606Jacqui Read treats communication as a design discipline in Communication Patterns: it has patterns and anti-patterns like any other. For a handoff that reads literally — “just write it all down in a document” is one of the anti-patterns.Book Cube · Communication Patterns Transparency matters especially when transferring people. Lara Hogan proposes a three-person conversation among the former manager, new manager, and employee. Goals, strengths, current feedback, and career expectations can move while the employee can clarify or disagree. This is safer than a secret dossier and preserves development continuity without requiring the successor to inherit every prior opinion.

A good ending begins with the order of communication. First speak to the person accountable for your role and transition, then to those whose decisions depend on the news, and—once aligned—to the team and key stakeholders. A manager should not learn through a rumor, and a team should not spend weeks inside a vacuum after a mysterious announcement. The sequence protects clarity of responsibility, not emotional control.

Separate four things in the first conversation: the decision, the proposed timeline, a brief honest reason, and a transition proposal. Do not turn it into a catalog of every grievance in the relationship. Do not leave false hope if the decision is final. Professional clarity does not require disclosing the next company or personal details you prefer to keep private.

Your position changes after notice. Before it, you might have initiated a long redesign. After it, large irreversible decisions should be made with the future owner and manager. That does not mean ceasing to work. Complete commitments, surface risks, support the team, and make necessary short-term decisions. Do not use the final weeks to force through everything the organization declined before.

Close relationships as deliberately as projects. Thank specific people for specific contributions. Correct the authorship of results so the team receives credit for its work. Give constructive feedback where there is an audience, an observation, and a possibility of action. An exit conversation is useful when it helps the organization learn. It is much less useful as a final safe place to unload concerns that were never named while change was possible.

Leaving well does not mean agreeing with everything, concealing harm, or promising permanent availability. It means preserving trust where direct and professional behavior remains possible. Former colleagues in technology become customers, candidates, partners, investors, and colleagues again. Research on organization–alumni relationships argues that value creation or destruction continues after formal employment ends. Reputation is built less by a graceful farewell message than by how someone crosses a difficult boundary.

It is convenient to draw the old and new jobs as separate stories. In practice they meet at one seam. Researching the next role is already part of its first 90 days. The quality of the old handover affects energy, reputation, and available attention in the new role. Unresolved promises follow. The desire to remain indispensable denies the successor room and prevents the departing person from moving psychologically.

Agree on post-departure boundaries before the final day. Is there a limited window for questions? Who collects them? Which channel is used? What qualifies as an emergency? Would formal consulting be needed and arranged if the company wants continuing support? Specific arrangements vary. Ambiguity reliably creates shadow leadership: former reports reach out privately, the successor looks temporary, and the former leader keeps making decisions without current context or accountability.

Larson warns executives about this pattern. Continuing the old one-to-ones, constantly mentoring the former team, or remaining visibly present can undermine the person who accepted the role. Helping once through an agreed request is not the same as remaining an informal center of power. Respect for the successor includes space for decisions you would have made differently.

The table is a management heuristic, not a standard. A long transition allows more deliberate work. A short one requires protecting the highest-cost commitments first. Three principles remain: the decision should rest on a diagnosis, responsibility should acquire a new owner, and relationships should acquire a clear new form.

What to be able to show yourself on the final day

Not a perfect archive and not proof that you were indispensable. Five things are enough. You can explain honestly how the move fits the desired trajectory. Every critical area has an owner and next action. People heard the news from the right participants and understand the transition. The company received useful feedback without concealed sabotage. Finally, you are able to stop performing the old role and direct attention to the next chapter.

Departure then becomes neither escape nor a performance of corporate loyalty, but a mature change of contract. 070707Stulberg and Magness separate the passion that feeds you from the passion that consumes you. A change of chapter is a rare moment to check which one drives your relationship with work — and to avoid carrying the old regime into the new role.Book Cube · The Passion Paradox Sometimes the best outcome is a new role inside the same organization. Sometimes it is an external move. Sometimes it is a pause. In every case, the work of the last 90 days is the same: make the choice deliberate, make the transfer operational, and leave the relationships intact enough for the past to support the next chapter rather than govern it.