# Onboarding Plan > Onboarding Plan - Skill: `ingridleiria/onboarding-plan` (Agent Skill) - Install (CLI): `npx skillmds@latest add ingridleiria/onboarding-plan` - Raw SKILL.md: https://api.skillmd.com/api/skills/ingridleiria/onboarding-plan/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: ingridleiria (https://skillmd.com/u/ingridleiria) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/ingridleiria/onboarding-plan --- # Onboarding Plan Most hires that fail were not bad hires. They were unplanned starts. The person arrived, was told to settle in and meet people, spent three weeks being generally available, and by the time anyone assessed them there was no agreed standard to assess against, only a set of impressions formed while they had nothing specific to do. The manager then reaches for a judgement that sounds like fit and is really an artefact of the absence of a plan. The cost falls in the wrong places. The organisation loses the months of output it hired for. The manager loses a quarter to a performance conversation that could have been a day 30 correction. The new person, often perfectly capable, loses standing that is hard to rebuild, because a reputation formed in the first six weeks survives contrary evidence for about a year. A written plan converts the first ninety days from a period of goodwill into a sequence of checkpoints with deliverables and dates. Its second function matters as much: it protects the hire from being busy without being useful, the state most new joiners drift into and the hardest to notice from the inside. ## When to use this, and when not to Use it for any new hire whose absence would be felt, from a first analyst to a chief financial officer. Use it for internal moves and promotions, where the plan is usually skipped on the reasoning that the person already knows the company, which is true and irrelevant: they do not know the new role, and the people who used to be their peers do not know how to treat them. Use it for a returning employee after a long absence, and for someone taking over from a departing owner, where the plan doubles as a handover schedule. Do not use it to define the role. The outcomes come from the scorecard in `hiring-scorecard-and-interview-kit` or an equivalent written statement of what the person is for; a plan built on an undefined role produces a busy ninety days ending in the argument hiring was supposed to settle. Do not use it as a performance improvement plan for someone already in post, which is a different instrument with different consequences and should never borrow this vocabulary. `operating-cadence-design` produces the meeting and reporting structure a new senior hire has to learn and often has to fix. `process-documentation-sop` is where the handover knowledge should end up when someone is leaving, so the incoming person reads a document rather than interviewing a memory. ## What you need before starting **The outcomes for the role, written.** Three to five, measurable, at twelve months. This is what the day 30, 60 and 90 reviews assess against, and without it every review becomes an impression. Missing: take them from the hiring scorecard. If there is no scorecard, spend an hour with the manager producing one before writing any plan, because the plan is otherwise a calendar. **The manager's real availability in the first month.** Onboarding is a manager's task that cannot be delegated to a document. Missing: find out now, not in week two. If the manager is away for three of the first four weeks, the plan changes shape: the buddy carries more, introductions move earlier, and the day 30 review is booked before the hire starts. **The organisational context.** Who the role works with, the systems, the documents that define how the place runs, the current priorities, the live problems. Missing: assemble a reading list from what exists, and where nothing written exists, tell the hire that plainly, or they will assume they failed to find it. **Access, equipment and accounts.** Missing: escalate before the start date, not on it. A hire who spends week one raising tickets has learned something no induction session will unteach. **The list of people to meet, with a reason for each.** Missing: build it with the manager and one long-tenured colleague, who will name the people the org chart does not. **What the hire inherits.** Live commitments, half-finished work, a relationship someone is worried about, a decision waiting on the vacancy. Missing: ask the departing person or whoever is covering, in writing. Inherited commitments discovered in week five are the most common cause of a missed day 60 deliverable. **Constraints.** Remote or on-site, time zones, visa or start-date limits, a notice period overlapping a critical event. Missing: assume the least convenient case and build a plan that survives it. ## The method 1. **Confirm the outcomes and get them agreed in writing before day one.** The hire, the manager, and where relevant the manager's manager. Twenty minutes, and it is the load-bearing step: every checkpoint resolves against it, and disagreement discovered at day 60 is expensive. 2. **Do the pre-start work, and treat it as part of the plan rather than as admin.** Accounts, equipment shipped and tested, the plan document itself sent a week ahead, a note from the manager, the first week's calendar populated with named introductions, a buddy assigned and briefed on what the role actually is. The judgement call is how much to send in advance: send the plan, the org context and the reading list, and do not send live work or anything confidential before the start date. 3. **Design days 1 to 30 for learning, with one deliverable at the end of it.** The goal is understanding the business, the people, and the current state of the role. Build the people list: fifteen to twenty-five for a senior role, eight to twelve for a junior one, each with why they matter and one question to ask them. The question is what turns a courtesy meeting into intelligence, and it also gives a nervous new joiner something to open with. Include peers, the manager's peers, adjacent teams, and for senior roles the board members, key customers or partners who will judge the function. Build the reading list: strategy documents, the operating cadence, the metrics, the last three board packs or their equivalent, the decision log, and the two or three documents that explain why things are as they are. Schedule observation of the recurring meetings and shadowing of the work the role will own. The day 30 deliverable is a two-page written read-back to the manager: what the hire now understands the business and the role to be, what surprised them, what they think the priorities should be. It surfaces misunderstandings while they are cheap, it is the first real evidence of judgement, and it is the first thing a manager can respond to concretely rather than encouragingly. 4. **Design days 31 to 60 for contribution, on something small and visible.** The rule for choosing it: real enough that a failure would be noticed, small enough that a failure would not be serious. A process fixed, an analysis delivered, a meeting taken over and run properly, a report rebuilt. The hire takes the first responsibilities from the outcomes, with the previous owner still available but no longer doing it. Relationships deepen with the five people the role depends on most. The manager collects feedback from those five and shares it, unedited on substance, at the day 60 review, because feedback that reaches the hire only through the manager's summary at day 90 arrives too late to act on. 5. **Design days 61 to 90 for ownership and a forward plan.** All the core responsibilities transferred. The hire writes their own ninety-day assessment: what is working, what is not, what they would change, and their plan for the next quarter with outcomes and dates. That document, rather than the manager's opinion, is the main artefact of the phase, and its quality is itself evidence. 6. **Book the three reviews in the calendar before day one.** Day 30, 60 and 90, each with a written input from the hire and a written output from the manager. The day 90 output is a decision in one of three forms: confirmed with the next quarter attached, on track with these named gaps and this support, or this is not working and here is what happens now. Silence at day 90 is the most common managerial failure, and the plan exists partly to make silence impossible to arrange. 7. **Match the check-in cadence to the role's ambiguity, not the person's seniority.** The default is a short daily check-in in week one, twice weekly to day 30, weekly after. A role with a clear queue needs less; a role that must invent its own agenda needs more, and senior hires in ambiguous roles are the group most often under-supported, because frequent check-ins feel like distrust. Say explicitly that the cadence follows the role and will reduce. Fix the shape of each check-in so it does not become a social call: what was learned or done, what is unclear, what is blocked, what is next. Fifteen minutes is enough when the shape is fixed. 8. **Assign a buddy and give them a job.** A peer, not the manager, meeting the hire weekly for the questions they will not ask upward: whether that meeting is always like that, whether it is normal nobody replied, who actually decides. Brief the buddy on the role, and tell them plainly they are not being asked to report on the hire, or the hire will work it out and the channel closes. 9. **Revise the plan at each review rather than treating it as fixed.** A plan that survives ninety days unchanged was not being used. What changes is usually the day 60 deliverable, because the day 30 read-back reveals a better candidate for it. 10. **Close the loop back to hiring.** At day 90, note which competency the interview process over-weighted and which it missed, and send that to whoever owns the scorecard. It takes ten minutes and it is the only feedback loop that improves hiring. ## The executive and Chief of Staff variant Senior hires rarely fail on capability. They fail on sequence: a change made in week three that was correct on the merits and cost them the relationship they needed to make the next five. Add three things. A structured listening tour in the first three weeks, with the same question set used with everyone, so answers are comparable and the exercise signals a method rather than a search for allies. A written diagnosis at day 30, delivered to the chief executive, judged on whether it names something uncomfortable that everyone already knows. And a first visible change by day 60, chosen for low risk and high visibility, because a senior hire who has changed nothing by then is assumed to have no view. The standing rule is to propose nothing structural before the day 30 diagnosis has been delivered and discussed. The exception is a live crisis, where waiting is worse than being wrong, and there you say plainly that you are acting before you understand the whole picture. For a Chief of Staff specifically, the day 30 deliverable is best made an operating cadence diagnosis, using `operating-cadence-design`. It maps the organisation, requires talking to everyone, produces something immediately useful, and gives the new Chief of Staff a legitimate reason to ask every leader what they think is broken. ## When you are the new hire Build the same plan from your own side and share it with your manager in week one. It works as a plan, and as evidence: a joiner who arrives with a written first ninety days has demonstrated something the interview could not. The differences are in emphasis. Write the questions for each introduction before the meeting and take notes you can search later, because the pattern across twenty conversations is the actual output and it is invisible without notes. Request documents rather than waiting to be given them. Choose your own early win and get it agreed rather than being assigned one. And hold your changes until the read-back has landed: the most common self-inflicted wound is the confident recommendation made in week two, based on a pattern from a previous organisation, which turns out to have been tried here the year before you arrived. ## Worked example **Situation.** A logistics software company of about 140 people hired a head of finance to replace a controller of six years who was leaving in five weeks. The company had raised a round eight months earlier and had promised investors a rolling thirteen-week cash forecast. Nothing was written down about the month-end close; the controller ran it from memory and a spreadsheet with tabs named after months of 2019. **Task.** Get the new head of finance to independent ownership of the close and a working cash forecast by day 90, without a gap in reporting during the five-week overlap. **Action.** The outcomes were taken from the hiring scorecard, and one was rewritten before day one. It had said "improve financial reporting", which fails the test that two people could disagree at twelve months about whether it happened. It became: month-end close completed within five working days by month four, and a thirteen-week cash forecast delivered weekly from month three with variance against the prior week explained. The pre-start work found two problems. The finance system licence was per-named-user and at its cap, which took nine days to resolve and would have consumed week one. And the departing controller's five-week overlap was in fact three, because of untaken leave nobody had counted. That changed the shape of the plan. The handover was front-loaded: the day 30 read-back was moved to day 21, and the first two weeks were built around `process-documentation-sop` sessions where the incoming head of finance documented the close as the controller performed it, rather than watching and taking notes. Writing the SOP while the expert was still present was the highest-value use of the overlap, and it produced a document rather than a set of recollections. The wrong turn: the original first-month deliverable was a review of the finance function's structure with a recommendation on hiring a second analyst. It was abandoned in week two, because the head of finance did not yet know why the close took eleven days, and a structural recommendation made before understanding that is an opinion about headcount dressed as analysis. It moved to day 75, by which point the answer had changed: the close was slow because of a manual reconciliation between two systems, and the recommendation became an integration rather than a hire. The people list was twenty-two names, including six customers and the investor director on the board, on the reasoning that a head of finance who has not spoken to the person who reads their numbers is at a disadvantage they will not recognise. The question for each was written in advance; the one that produced most was asked of the four regional operations leads, what number do you not trust, which surfaced a revenue recognition inconsistency nobody in finance had known about. Day 60 was the first cash forecast, delivered to the chief executive and the board's finance contact. It was late by four days and the variance explanation in the first version was mechanical rather than causal, which the manager said plainly at the review rather than saving for day 90. **Result.** At day 90 the close ran in seven days against a target of five by month four, on trajectory. The forecast was weekly and its variance explanations were causal by the fourth iteration. The written assessment confirmed the hire and named one gap, a reluctance to push back on the chief executive's revenue assumptions, with a support plan of joint sessions before each board pack. The close SOP written in the first fortnight was used twice more that year: once during the head of finance's leave, and once to onboard an analyst in three days rather than three weeks. The feedback sent back to hiring was that the process had assessed technical accounting thoroughly and willingness to contradict a senior person not at all. That competency went into the next finance scorecard. ### A second scenario, where it goes differently The same company, a year later, hiring three implementation specialists into an existing team of nine, all starting on the same Monday, with a manager who has onboarded seven people into the identical role. Nearly everything changes. The role is understood, the work has a queue, the systems are documented, and there is a cohort to learn from. The people list drops from twenty-two to nine, the same nine for all three. The day 30 deliverable becomes a much smaller artefact: one implementation completed end to end with a reviewer, rather than a written diagnosis of the business. Check-ins run daily for two weeks then weekly, delivered to the cohort together as well as individually. The buddy is a peer who joined four months earlier and remembers what was confusing. What does not change is the dated reviews with written outcomes, and here they carry more weight, because with three simultaneous hires the natural failure is to assess them against each other rather than against the standard. Each is reviewed against the outcomes alone, in writing, before any comparison, and the comparison is used only to decide who needs what support. What changed is the amount of discovery the role requires. The ambiguity of the role, not the seniority of the person, is what sets how much of this plan is needed. ## Output One document, sent before day one, revised at each review. ``` ONBOARDING PLAN Name: [ ] Role: [ ] Start date: [ ] Manager: [ ] Buddy: [ ] Reviews booked: Day 30 [date] Day 60 [date] Day 90 [date] OUTCOMES AT TWELVE MONTHS (agreed by [names] on [date]) 1. 2. 3. BEFORE DAY ONE | Item | Owner | Due | Done | | Accounts and access | | | | | Equipment | | | | | Week one calendar | | | | | Plan sent | | | | DAYS 1 TO 30: LEARN People to meet | Person | Role | Why they matter | The question to ask | Booked | To read | Document | Where | Why | Deliverable by day 30: [the read-back, with its audience] DAYS 31 TO 60: CONTRIBUTE Responsibilities taken on: [ ] Handover from: [ ] Deliverable by day 60: [with its real audience] Feedback collected from: [five names] DAYS 61 TO 90: OWN Remaining responsibilities transferred: [ ] Deliverable by day 90: the hire's own assessment and next-quarter plan CHECK-INS [Cadence, and the fixed four questions] ``` The review template, used three times: | Outcome | Evidence so far | On track / at risk / off track | What changes before the next review | Owner | Ending in one line: confirmed, on track with named gaps and support, or not working with what happens next. ## Failure modes **The plan that is a calendar.** Recognise it because every entry is a meeting and none is a deliverable. Fix by requiring one artefact per phase with a named audience outside the reporting line; work nobody outside it consumes establishes nothing. **No plan for a senior hire, on the grounds that they will work it out.** The most expensive version, because senior hires get the most ambiguity and the least structure at exactly the point they have the least context. Recognise it in the sentence "they are experienced, they will find their way". **Access requested on day one.** Recognise it in an empty first week because the person cannot log in. Fix by giving the pre-start checklist to someone with authority, dated two weeks before the start. **Introductions with no question.** Recognise them because the hire cannot say afterwards what they learned. Fix by writing one question per person into the plan before the meeting. **Feedback held until day 90.** Recognise it when the day 90 review contains a surprise. Any concern that appears there and was not raised at day 60 is a management failure, whatever it says about the hire. **The plan abandoned in week three.** Recognise it because the day 30 review was never booked. Fix by booking all three reviews before the start date, in the calendar, with the manager's manager copied on the day 90 one. **The hire who is fully occupied and producing nothing.** Recognise it when check-in answers are all activity and no output for two weeks running. Fix by naming the next deliverable and clearing something to make room. ## Edge cases **No scorecard and no agreed outcomes.** Do not build the plan first. Spend an hour producing three outcomes with the manager, get them in writing, then build. A plan on undefined outcomes produces a busy quarter and a review nobody can resolve. **The role is new and nobody knows what good looks like.** Make the day 30 deliverable the definition itself: the hire proposes the outcomes, the manager agrees or amends. Then run days 31 to 90 against what was agreed. State this as the plan rather than letting it be discovered. **A start date with no manager, because the manager is leaving or absent.** Assign an interim owner with real authority, not a caretaker, and book the day 30 review with whoever will be permanent. Onboarding without a named owner reverts to nobody within two weeks. **Replacing a well-liked predecessor.** Add an early conversation with each of the five people closest to the previous holder, and change nothing visible before day 60 unless it is broken. The technical handover is rarely the hard part here. **A very short runway, where the hire must deliver in week two.** Compress but do not skip: a day 10 read-back, a people list of the five who matter most, and a written note of what is being deferred, so a later problem is a known trade rather than a surprise. **The plan reveals in week four that the hire is wrong for the role.** Act at day 30 rather than waiting for day 90. The early version of that conversation is the kindest and the cheapest, and the dated reviews are what make it possible to have. ## Quality bar - The outcomes are written, measurable, and agreed by the hire and the manager before day one. - Access, equipment and the week one calendar were ready before the start date. - Every phase has one deliverable with a named audience outside the reporting line. - The people list names individuals with a reason and a question each, not categories. - Day 30, 60 and 90 reviews are booked as dated meetings before day one, each with a written input and a written output. - The day 90 output is a decision in one of three stated forms, never silence. - The check-in cadence has a fixed shape and a stated end point. - Anything raised at day 90 was already raised at day 60. ## Adapting this to your context The 30-60-90 shape, the people-list sizes and the check-in cadence come from fast-growing commercial companies of fifty to five hundred people, where a joiner can be productive within a quarter. - **Ninety days as the horizon.** Regulated, clinical and licensed roles have a credentialing period that owns the first phase. Start the ninety days when it clears, and say so in the plan. - **Fifteen to twenty-five introductions for a senior role.** Add a layer in a matrixed or multi-country organisation, where the people who can block the role sit two functions away. Cut it hard where the role has a defined queue. - **A visible change by day 60 for an executive.** In a partnership, a public body or a long-tenured team the same move reads as arrogance. There the day 60 artefact is the diagnosis, discussed with the people it affects. - **The buddy who is not reporting on the hire.** Where a works council, union agreement or probation framework governs the relationship, check what may be recorded before promising confidentiality. - **What not to change.** The outcomes are written and agreed before day one, and the day 90 output is a decision in one of three stated forms. Silence at day 90 is the failure this prevents. ## Related skills `hiring-scorecard-and-interview-kit` produces the outcomes this plan runs against, and receives the day 90 feedback on what the interview process missed. `operating-cadence-design` is the natural day 30 deliverable for a Chief of Staff or operations hire, and what a senior joiner usually has to learn before changing anything. `process-documentation-sop` captures a departing person's knowledge so the incoming one reads rather than interviews, which is the highest-value use of a short overlap. `weekly-status-update` is the reporting habit to adopt from week one. `decision-memo` is the format for the first structural recommendation, which should not appear before the day 30 read-back. `principal-simulator` helps a senior joiner calibrate to a new chief executive.