How to Write a Hiring Plan: The Document, the Fields, and the Approval Path

What information does each row in the plan need?

Name the outcome rather than the title. A row that reads senior backend engineer tells a reader nothing about why the seat exists. A row that reads owns the billing rewrite so that enterprise contracts can be invoiced without manual work can be argued about, prioritized, and cancelled on evidence. Titles come later, when you write the description.

Every row needs a single named owner who will run the search and manage the person, an explicit statement of whether the seat is incremental or a backfill, the earliest month it may open, and its current approval state. Approval state matters more than it sounds: distinguishing approved, conditionally approved, and requested prevents the common situation where three people believe a role was signed off and nobody can point to when.

Add two fields that most plans omit. First, the condition under which the row does not happen, expressed as something observable. Second, what happens instead if it does not, which is usually that a named person absorbs the work or a specific deliverable slips. Writing the alternative forces the plan to be honest about whether the seat is genuinely necessary or merely desirable.

Who should own the plan and who approves a new row?

One person maintains the document, and it should be whoever also maintains the cash model, because the two are the same conversation seen from different angles. Distributing maintenance across function leads produces several versions that diverge within weeks, and the version people quote in meetings will be whichever one supports the point being made.

Approval authority should be explicit and short. State who can approve a row, who must be consulted, and what threshold escalates. In a small company that is usually a founder with the finance owner consulted, and stating it in one sentence removes an entire category of confusion about whether a hiring manager mentioning a role in a meeting constitutes approval.

Separate approving the row from approving the offer. A row being in the plan means a search may open at a certain time; it does not commit you to any particular candidate at any particular number. Keeping those decisions distinct means a hiring manager can proceed without checking in constantly, while the compensation decision still gets its own moment of scrutiny.

How do I write the request that opens a role?

Require a short written case, kept to a page. It should say what work is not getting done today and what evidence supports that, what the person will own in their first six months expressed as outcomes, who else was considered as an internal alternative, and what happens if the role is deferred by one quarter. A page is enough and a page is a genuine filter, because the roles that cannot be justified in a page are usually the ones that were requested out of frustration.

Require the alternatives to be addressed rather than mentioned. Could the work be automated, absorbed by reprioritizing, done by a contractor for a defined period, or handled by moving someone internally? Most requests skip this section, and reading a serious attempt at it changes the decision more often than any other part of the document.

Then have the requester name the measure that would show the hire worked. Not a job description, a measure: the queue that shrinks, the deliverable that ships, the revenue that appears, the response time that improves. Six months later that sentence is how you evaluate whether the plan is working, and without it every hire looks successful because someone is busy.

What cadence keeps the plan accurate?

Review monthly for thirty minutes against actuals: which searches opened, which slipped, which offers were accepted, which start dates moved. This is a status meeting and should not turn into a discussion of strategy. Its purpose is to keep the document matching reality so that the quarterly conversation has a trustworthy baseline.

Review the composition of the plan quarterly and with more scope. Which rows are still justified given what has changed, which have been overtaken by a product decision, which functions are now under-resourced relative to where the constraint has moved. Expect to remove rows here. A plan that only ever grows is a document nobody is willing to argue with.

Version it and keep the old versions. Being able to look at what you believed two quarters ago is the fastest way to improve at planning, because it shows you the specific direction in which you are consistently wrong. Most founders discover they systematically underestimate elapsed time to fill and systematically overestimate how quickly a new person becomes productive.

How do I decline a request that is not in the plan?

Route it through the same written case as any other row rather than answering in the moment. This converts an interpersonal disagreement into a document comparison, and it lets you say that the request is being considered against the same standard as everything else rather than being dismissed. Most urgent asks either strengthen considerably when written down or quietly evaporate.

When you decline, decline the timing rather than the need where that is the honest answer, and say what would change it. A request refused with a stated condition, such as revisiting after a specific event or once a named measure moves, keeps the requester engaged in reaching that condition. A flat no with no explanation gets re-litigated every month.

Track declined requests in the same document. A role requested three times across three quarters by three different people is signal, not noise, and it is invisible if each refusal is a separate conversation that leaves no record.

How should backfills be treated differently from new seats?

Do not automatically approve a backfill. A departure is the one moment when the shape of a role can be reconsidered at no political cost, because nobody is being asked to change. Ask whether the same role should be refilled, whether it should be split or merged, whether the level is right, and whether the work still needs doing at all.

Mark backfills distinctly in the plan, because they behave differently financially. A backfill is roughly cash-neutral against the departing person's cost, subject to any change in level and any overlap period, while an incremental seat adds to burn. Mixing them in a single count makes the headcount number look alarming or reassuring for reasons unrelated to spend.

Set a short time limit on the reconsideration so it does not become a delay tactic. A week to decide whether the role changes shape is enough; a month of deliberation while the team absorbs the gap costs more than the flexibility is worth.

What are the common failure modes in hiring plan documents?

The first is a plan written entirely in titles, which makes it impossible to challenge any row on its merits. If every entry is a job title and a quarter, the only available discussion is whether the total is too high, which is the least useful version of the conversation.

The second is the plan that lives in a deck. Decks are built for a moment and are never updated afterward, so within two months everyone is working from a stale slide. Keep the plan in a document or sheet that is edited in place, and let the deck quote it rather than replace it.

The third is a plan with no owner for individual rows. When a search stalls and nobody is accountable for it, the row sits open for months and quietly makes the whole plan look unreliable. Every row needs a name, and that name should be the person who reports on it at the monthly review, not a function.

What to take away

• ▸Write each row as the outcome the seat produces, not as a job title with a quarter beside it.
• ▸Give every row a named owner, an approval state, a backfill or incremental flag, and an earliest open month.
• ▸Require a one-page written case that addresses automation, reprioritization, contract, and internal moves.
• ▸Keep approving a row and approving an offer as two separate decisions.
• ▸Review monthly against actuals for accuracy and quarterly for composition, and expect to remove rows.
• ▸Never auto-approve a backfill; a departure is the cheapest moment to reshape a role.