The First Finance Hire: Structure the Seat First

The first real finance hire in a 10 to 75 person company fails at a rate that would be a scandal in any other seat, and the failure is nearly always designed in before anybody is interviewed. Somebody says the word “controller” because it is the word they know. The posting goes up with a title at the top and a paragraph of duties underneath that nobody derived from the actual work. A capable, expensive person is hired, arrives, finds no process to step into, and spends six months building what should have existed before they were offered the job. At month nine the owner concludes the hire was wrong. The hire was fine. The seat did not exist.

So the whole of the work is up front. Before a job is posted, someone has to count the work that actually exists, decide which shape of seat that work makes, and put in place the small number of things a new person cannot reasonably be asked to invent. That is a few weeks of unglamorous effort by people already in the building, and it is the difference between a hire and an experiment.

Here is the position, stated early because it is the whole argument. If the process does not exist before the hire, you are not making a finance hire. You are commissioning a build project and attaching a salary to it. That is a legitimate thing to do. It is a completely different job, it attracts a different candidate, it takes longer, and it has to be said out loud in the posting and in the first interview. What is not legitimate is doing it by accident and then judging the person against a steady state that was never running.

What follows is one hire walked end to end: deciding what the seat is, writing the job, the things that go in before day one, three closes of ramp, and the test at six months.

Count the work before you name the seat

Nobody can specify this role from a title, because the same title covers four different jobs depending on what the company has already automated, already outsourced and already broken. The only reliable starting point is an inventory of the work as it is performed today, including the work currently done badly and the work currently not done at all.

Take twelve months. List every recurring finance task. For each one, record who performs it now, how often, how long it takes in a real month rather than in memory, and what evidence it produces. Then assign it to one of four bands.

Band	What sits in it
Capture	Transaction entry, AP entry and payment runs, customer invoicing, collections calls, payroll input, expense reports, bank and card reconciliation
Close	Sub-ledger agreement, balance sheet reconciliations and the schedules behind them, recurring journals, accrual calculation, the reporting pack, variance commentary
Judgment	Revenue cut-off and policy application, estimates and their basis, unusual transactions, presentation decisions, the year-end file and the practitioner relationship
Forward	Cash forecasting, budget build, pricing and margin analysis, financing and covenant work, board or shareholder reporting

The inventory itself is a table you can hand to a candidate. Build it in this shape.

Task	Band	Frequency	Performed today by	Evidence it produces
Bank and credit card reconciliation	Capture	Monthly	Bookkeeper	Reconciliation with statement attached
AP entry and weekly payment run	Capture	Weekly	Bookkeeper and owner	Approved batch, payment file
Payroll input and journal	Capture	Per pay period	Office manager	Register agreed to the journal
Deferred revenue schedule	Close	Monthly	Nobody, currently	A schedule that does not yet exist
Accrual review and posting	Judgment	Monthly	External accountant, at year end only	Adjusting entries, twelve months late
Reporting pack and commentary	Close	Monthly	Owner, from a spreadsheet	A P&L nobody reconciles
Sales tax return preparation	Capture	Per filing cycle	Bookkeeper	Filed return, reconciled payable
Cash forecast	Forward	Never	Nobody	The bank balance, read on a Friday

Leave the hours column blank in the template and fill it from an observed month. Estimated hours run low on capture work and high on everything else, consistently, in every company where we have measured both.

The shape matters more than the total

The total hours tell you whether there is a seat at all. The distribution across the four bands tells you which seat it is, and that second reading is the one people skip.

If Capture and Close dominate and Judgment is a handful of hours a month, the seat is a strong bookkeeper or an accountant, and the judgment hours have to be assigned somewhere else in writing. Hiring a controller into that inventory produces a person who does reconciliations for four months, gets bored, starts a systems project nobody asked for, and leaves in month ten. That is the single most common version of this failure and it is entirely self-inflicted, because the inventory said what the job was before anyone read it.

If Close and Judgment are both substantial, and the pack is currently late or issued and unread, the seat is a controller and the title is correct for once. Two conditions apply. There has to be enough Capture work for someone else, because a controller who is also the only data entry clerk will do the data entry first, every month, and the review work will be the part that slips. And the company has to accept that the controller will say no to things, which is most of what you are paying for.

If Forward dominates, this is not this hire at all. Forecasting, pricing and financing work sits above a close, and if the close underneath is not running then the forecast is a well-formatted opinion. Fix the order.

There is a fourth reading and it is the awkward one. If the total does not add up to a full-time load, the honest conclusion is that there is no seat yet. See below, because that case has a real answer and it is not a smaller version of this hire.

Writing the job

The posting is the inventory, restated for a reader who does not work here.

Structure it in four parts, in this order.

What you will own in month one. The actual tasks, by name, with the day of the close calendar each one falls on. This is the section that gets you the right applicants and loses you the wrong ones, which is the entire purpose of a posting.
What you will own by month six. The items on the inventory currently performed by the owner, the office manager or nobody, and the date they transfer.
What this role does not include. The bands going elsewhere, and to whom. A candidate is also making a decision, and the ones worth hiring read this section hardest.
The conditions. Entity count, transaction volumes, the systems by name, whether there is inventory, whether there is a lender covenant, whether year-end brings a practitioner in and at what level of engagement. These are the facts that make a job a different job, and leaving them out is how you interview six people who are wrong for you.

Three rules for the writing itself.

Delete “other duties as assigned.” In a first finance hire that line is not boilerplate. It is the trapdoor under which every undefined thing in the company eventually gets loaded, and both parties will point at it in the month nine conversation.

Do not require a professional designation unless a task on the inventory needs one. Requiring it narrows the field and raises the price for reasons that are frequently about the buyer’s own comfort rather than about the work. Decide by reading the Judgment band. If the year-end file goes to an external firm and the technical positions are theirs, you are paying a premium for something you are already buying elsewhere.

Publish the close calendar, or admit there is not one. If the answer to “when does the month close” is “the third week, usually”, say so in the posting. A candidate who accepts a role knowing that is a candidate who has priced the work correctly.

The interview that actually tests something

Two exercises, and they are worth more than the rest of the process combined.

Hand the candidate a redacted trial balance and the most recent bank reconciliation, give them twenty minutes, and ask what is wrong. The specific defects matter far less than the route they take to them. Watch whether they open the balance sheet before the income statement, whether they ask what the suspense account is for, and whether they say “I would need to see the sub-ledger” rather than guessing. Anybody who has run a close does this the same way and cannot fake not doing it.

Then ask about the last time they were told to book something they disagreed with. What was it, what did they do, and what happened afterwards. The answer you want is a specific transaction, a conversation that was uncomfortable, and an outcome that occasionally went against them. A candidate with no such story either has not held the pen or will not hold it here.

What has to exist before day one

This is the list, and none of it requires the new hire. Every item is work the company can do before it posts the job, using people who are already there and, where it is genuinely beyond them, a few weeks of outside help. If the company cannot get through this list, hiring faster does not fix it. The list is the finding.

A close calendar with dates on it, even a bad one. It can be twenty days long and half wrong. What matters is that a sequence exists on paper, because the new hire’s job in month one is to run it and mark it up, and you cannot mark up an absence.
A chart of accounts nobody intends to change for six months. If it needs rebuilding, rebuild it before the hire starts, not after. A restructure during the ramp destroys the comparatives the new person is using to learn the business, at exactly the moment they are least able to tell a real variance from a mapping change.
Every balance sheet account with a stated position. Reconciled with evidence, supported by an internal schedule, or explicitly known to be unreconciled with a note saying so. You are allowed to hand over a mess. You are not allowed to hand over a mess that nobody has inventoried, because then the new hire spends their first two months discovering it one account at a time and reporting it as bad news, which sets the tone of the relationship for a year.
System access, in the company’s name, provisioned and tested before the start date. Access granted is not access that works. Somebody logs in as the new user and confirms it, the week before, not on the morning.
A named approver for judgment items, who is not the new hire. In month one the hire is not the last word on an accrual or a revenue cut-off. If the fallback is the owner, say that. If it is an outside reviewer, arrange it before the start date and tell the hire on day one who to call.
Twelve months of closes in one place. Working papers, issued packs, the trial balance history, the journal listing, last year’s file and the practitioner’s adjusting entries. If any of it lives in email, it moves first.
A written statement of what the hire is not expected to fix in the first six months. One page. It is the most generous thing you can give a new finance person and almost nobody does it, because writing it requires the owner to admit which problems are being deferred on purpose.

Item seven is also a control on yourself. A list of deferred problems that runs to two pages is a company telling itself something.

When the honest answer is not yet

Three situations where the right decision is to delay the hire. The first is work this firm does. The second is not, and the third only sometimes is. That distinction is worth marking, because this subject sits close to something we sell and a piece that concludes with a reason to call us would be worth less than nothing to the person reading it.

The ledger is not closeable. Suspense has been growing, the last year-end produced a long list of adjusting entries, or the person who understood the file left. Hiring into that means the first six months are remediation, performed by somebody recruited to run a steady state and measured against a standard that cannot be met while the underlying data is wrong. Stabilise first, as its own piece of work with its own end point, then hire into the result. Remediation is a defined engagement here and it is deliberately separate from the recurring close for exactly this reason.

The work does not add up to a seat. This is the case where commercial honesty costs something, so here it is plainly, and it does not point at us. The answer is a competent part-time bookkeeper, a defined monthly review by somebody experienced, and an evening of the owner’s own time each month spent reading the pack properly. A full-time person hired into a part-time inventory will fill the remaining hours, and what they fill it with is process elaboration, new reports and a systems project, none of which anyone asked for and all of which then need maintaining. Buy the hours you need. Add the seat when the inventory says there is one.

Nobody in the company can tell a good close from a plausible one. This is the genuinely dangerous case, because it looks fine for a while. The hire produces a pack every month, the numbers are internally consistent, and nobody is capable of checking them. There is no review layer, no correction, and no feedback, and the first material error is discovered by the bank or at year-end. If that describes your company, you need a review layer above the hire before you need the hire, and that is the specific gap our fractional finance lead engagement is built to sit in. It is worth saying that the same gap is closed by any experienced reviewer you trust, including your own external accountant if they will take the work on a monthly rather than annual basis. What is not an option is nobody.

If you currently outsource the function and are hiring to bring it back in, the transfer itself has a different shape and a different set of failure modes, and it is covered separately in the finance handover walkthrough.

Ninety days, three closes

The ramp is measured in closes, not in weeks, because the close is the only event that exercises the whole function. Three of them fit in ninety days and each one has a different rule.

Close	The hire’s role	The rule	What exists at the end
Close 1	Observes, takes notes, touches nothing	Change nothing	The hire’s own written description of the process as they found it, and their defect list
Close 2	Runs it, hands on keyboard, with the current owner of the work beside them	Still change nothing	A completed close on the existing calendar, with the defect list revised by contact with reality
Close 3	Runs it and issues the pack alone	Fix exactly one thing	An issued pack on the hire’s own sign-off, and one defect closed properly

Insist on the deliverable at the end of close one, which is a document rather than a close: what they observed, in their words, and what they think is wrong with it. That is the most valuable diagnostic your company will ever receive about its own finance function, and you can only get it once, because by week eight the process looks normal to them and the questions stop.

The rule that gets broken is the second one. A capable hire will want to fix things in close two, and the instinct is a good sign. Hold it anyway. A process changed before it is understood destroys the evidence of why it was that way, and the reason is often a real constraint, a customer requirement or an old dispute that nobody thought to mention. One fix at close three, chosen by them, documented, is worth more than five improvised in week three.

Between the closes, schedule the work that will not appear on its own inside ninety days: the sales tax filing cycle, the payroll remittance rhythm and the annual sequence behind it, the year-end file and the practitioner relationship, covenant reporting if there is a lender, insurance and lease renewals. Each gets a booked session with a named person, in the calendar, before the start date. Anything not booked will be discovered by the new hire at the worst moment, on their own, and they will assume it was a test.

How the owner has to behave

Two habits kill a good hire in this seat, and both are the owner’s.

The first is routing around them. It is faster to approve the invoice yourself, faster to tell the bookkeeper directly, faster to ask the external accountant the question. Every one of those is individually reasonable and the cumulative effect is that by month four the new person owns the ledger but not a single decision, which is not the job they took. Route everything through the seat, including the things you could do in thirty seconds, especially in the first ninety days.

The second is judging the pack instead of reading it. A new finance hire’s first two packs will contain something that makes the business look worse than you believed it to be, because previously nothing was accrued and now it is. Reacting to that as a performance problem teaches a very fast lesson about what this company wants its numbers to say, and you will not get an unwelcome number again for a long time.

The six month test

At six months, “working” is falsifiable. Test it against this and not against whether the person seems busy.

The pack has been issued on the calendar date for three consecutive months, and you did not have to ask where it was.
You have stopped checking the bank balance to find out how the month went.
The hire has refused something. A transaction they would not book the way it was requested, a payment they held, a date they would not move, and it stuck.
Somebody outside finance has changed a decision because of a number in the pack. A price, a hire, a supplier, a service line.
Items from the close one defect list have been closed, and closed with documentation rather than with the hire’s own memory.
The hire has produced something nobody asked for, and it was useful.
If a year-end fell inside the window, the file went out and the practitioner’s questions were about judgment rather than about reconstruction.

Items three and four are the ones that separate a finance function from a bookkeeping service. Everything above them can be true of a person who is simply doing the tasks accurately, which is worth having and is not what you hired.

Three things get mistaken for this test and none of them belong in it. A fast close proves nothing on its own, because a close can be fast precisely because nothing is being reviewed. Volume of output proves less. And a finance hire who has made nobody uncomfortable in six months has probably not started yet.

If at six months the only thing that has changed is that someone else is performing the same work at a higher salary, the seat was never designed. It was filled.

The First Finance Hire, and Why the Seat Has to Exist Before the Person Does