Handing the Finance Function Back to an In-House Hire
Every outsourced finance engagement should be built so it can end cleanly. This is the handover we run, including the parts that make the outgoing provider easy to replace.
An outsourced finance function is a stage, not a destination. Somewhere between 25 and 60 people, most incorporated Canadian companies reach a point where the work is steady enough, and close enough to the operating decisions, that it belongs to an employee. When that happens, the handover should take about ninety days and should not be difficult.
It is usually difficult anyway, and the reason is almost never bad faith. It is that the engagement was never built to be handed over. Files live in the provider’s tenant. Software is on the provider’s subscription. The recurring journal that adjusts a specific accrual exists only in one person’s habit. None of that is malicious. It is just what happens when nobody decides at the start that the company owns its own finance function.
So the handover starts on day one of the engagement, not on day one of the handover.
The ownership rules, set at the start
These are the rules that make a clean exit possible. If you are reading this in the middle of an existing arrangement, this list is also a useful audit of how portable your current setup is.
| Rule | Why it matters at handover |
|---|---|
| Every accounting system, document store and payroll platform is subscribed and paid for by the company, in the company’s name | A subscription owned by the provider walks out with them |
| The company is the primary administrator of every system, always | Otherwise the exit depends on someone else granting the exit |
| All working files live in the company’s own document storage, not the provider’s | Working papers are useful only if they are where you can reach them |
| All data must be exportable in an open format | A schedule locked inside a proprietary tool is not a deliverable |
| Every recurring process is written down, not remembered | Undocumented habit is the definition of lock-in |
| Naming conventions and folder structure are the company’s standard | So a new person can navigate without a tour |
| No process depends on a single individual’s personal login or personal device | Personal accounts cannot be transferred |
None of these cost anything. They just have to be decided rather than defaulted into.
Deciding what role to hire
Before handover planning, be honest about which job you are filling. The most common handover failure is hiring at one level and expecting the output of another.
| Role | Handles | Does not handle | Handover pattern |
|---|---|---|---|
| Bookkeeper | Transaction capture, AP, AR, payroll input, bank reconciliation, routine sub-ledger work | Judgment accruals, revenue cut-off decisions, statement review, technical accounting | Provider hands over the capture and reconciliation blocks, retains review and judgment |
| Senior bookkeeper or accountant | All of the above, plus most of the close, most schedules, first-pass variance analysis | Technical accounting positions, framework decisions, statement presentation | Provider hands over the close, retains review, technical work and the year-end file |
| Controller | The complete close, review, reporting, controls, the year-end file, the practitioner relationship | Capital structure, financing strategy, board-level work | Full handover, provider exits |
| Finance director or fractional CFO | Everything above plus forecasting, financing, board reporting | Daily transaction work | Not a replacement for the close, a layer above it |
A common and sensible pattern at 25 to 40 people is to hire a strong bookkeeper or accountant and keep a light review layer for a period, then hand over the review layer once the person has grown into it. That is a valid outcome and it should be stated openly rather than allowed to become a permanent dependency by drift.
The handover package
The package is assembled before the new hire’s first day, not during their first month. It is a folder in the company’s own storage, and it contains these things.
Documentation
| Document | Contents |
|---|---|
| Close calendar | Every task, its day, its owner, and what done looks like |
| Chart of accounts guide | The numbering scheme, what belongs in each block, and the rule for adding an account |
| Recurring journal entry register | Every standing entry: what it is, when it posts, how the amount is derived, which schedule supports it, and when it ends |
| Reconciliation index | Every balance sheet account, what it reconciles to, where the evidence lives |
| Revenue recognition memo | Arrangement types, when revenue is recognized, how deposits and multi-period work are handled |
| Accrual and estimate memo | Every judgment estimate, the basis used, and who approves it |
| Systems map | Every system, what it does, what flows into what, and where each integration writes |
| Payroll runbook | Pay calendar, the steps in a run, benefit and deduction handling, remittance process, year-end sequence |
| Sales tax runbook | How the return is prepared, which reports are run, how the payable is reconciled |
| Vendor and customer master notes | Payment terms, quirks, who approves what, anyone on hold and why |
| Banking and authority matrix | Who can approve, initiate and release payments, at what limits, and the dual-control rules |
| Year-end and PBC file | Last year’s full working paper file, the practitioner’s adjusting entries, and the PBC list |
| Open items register | Everything unresolved, with an owner and a date |
| Known issues memo | The honest list: what is fragile, what is a workaround, what should be fixed and was not |
That last document is the one that separates a real handover from a performance. Every finance function has workarounds. Writing them down is a gift to the person taking over, and hiding them is a trap they will step in during their third week.
The last twelve months of substance
Twelve months of close binders, complete and in order. Twelve months of issued reporting packages. The trial balance history. The full journal entry listing. The intercompany reconciliation history if there is a second entity. The budget and the actual variance commentary. If any of that only exists in email, it moves into the folder before handover starts.
The access register
Access is the part that goes wrong quietly, because it fails weeks later when something expires or an automated process stops running under a departed identity.
Build a register listing every system, every credential, and every automated process running under an identity. Then work through it.
| System | Company-owned? | Provider access level | Transfer action | Verified |
|---|---|---|---|---|
| Accounting system | Yes | Admin | Add new hire as admin, remove provider, verify company retains primary admin | |
| Banking portal | Yes | View and prepare only | Remove provider, confirm approval limits and dual control still function | |
| Payroll platform | Yes | Admin | Transfer, and confirm the year-end filing contact is a company identity | |
| Government and tax portals | Yes | Delegated representative | Remove the delegation, add the new hire, confirm authority is company-held | |
| Document storage | Yes | Editor | Confirm ownership of every folder sits with the company, not with a provider account | |
| Expense tool | Yes | Admin | Transfer, and check approval workflows do not route to a departing identity | |
| Merchant and payment platforms | Yes | View | Remove provider | |
| Scheduled reports and automations | Yes | Runs under identity | Reassign every one, then confirm each fires on schedule |
Two rules for the transfer itself. Credentials never move as plain text in an email or a chat message, they move through the company’s password manager with ownership reassigned, or through each system’s own invitation flow. And access is removed only after the new hire has confirmed working access, never before, so nobody is locked out of a live payroll run.
The verification column is not decorative. Access that was granted is not the same as access that works, and the difference typically shows up on the worst possible day.
The four-close taper
Ninety days, four month-end closes, and a deliberate reduction in involvement at each one. This is the whole method.
| Close | Provider role | New hire role | Test at the end |
|---|---|---|---|
| Close 1, shadow | Runs the close normally, narrating every step, screen-shared | Watches, takes notes, asks anything | The new hire can describe the sequence and name the outputs |
| Close 2, supervised | Sits alongside, corrects in real time | Runs it, hands on keyboard | The close completes on schedule with the provider intervening only on judgment items |
| Close 3, observed | Available, does not touch the file. Reviews only after issue | Runs it end to end, including the reporting package | Provider’s review finds nothing material that the new hire had not already flagged |
| Close 4, solo | Not involved. Reads the issued package as an outsider would | Runs and issues it | The package is issued on the calendar date, on the new hire’s own sign-off |
The taper works because it moves in one direction and the steps are named in advance. Without the structure, handover becomes an indefinite period during which both parties do a bit of everything and neither is accountable, which is comfortable for exactly as long as nothing goes wrong.
Between closes, cover the non-monthly work explicitly, because it will not appear naturally in a ninety-day window. Sales tax filing cycles, payroll year-end, the annual insurance renewal, covenant reporting, the year-end file, the practitioner relationship. Each gets a scheduled session with the runbook open.
The exit test
The handover is complete when the fourth close is issued without provider involvement and all of the following are true. Not before.
- The close was issued on the calendar date, not late.
- Every balance sheet account is reconciled with evidence in the binder, by the new hire.
- The new hire wrote the variance commentary and it explains causes.
- Every recurring journal entry posted correctly, on its own, without a reminder.
- Every scheduled report and automation ran under a company identity.
- No system, file, schedule or process still requires the provider’s access to function.
- The open items register has an owner for every line, and none of those owners is the provider.
- The new hire has answered a question from the owner or GM that was not in the documentation, correctly, from their own understanding.
Item eight matters more than the other seven. The first seven prove that the process transferred. Item eight is the only one that proves the understanding did.
The safety net, and its expiry date
Offer a short, defined period after the exit test: thirty days, questions answered, nothing performed. Not a retainer, not access, not a standing engagement. A named person, a defined window, and an end date in writing.
Be skeptical of anything that outlives it. If a company still needs the outgoing provider six months later, the handover did not finish and both parties agreed not to notice. The honest response is to name what did not transfer and finish that specific piece, on a defined scope, rather than converting an unfinished handover into an indefinite arrangement.
The two things genuinely worth keeping separate from the in-house role are the practitioner who reports on your financial statements, who must be independent of you, and specialist technical work that arises rarely enough that nobody should be employed to do it. Those are legitimate outside relationships. Routine monthly work is not, once you have hired someone to do it.
The sign-off sheet
End the engagement with one page, signed by both sides. It records the date of the final close handed over, confirmation that every item on the package inventory was delivered, confirmation that every line on the access register was transferred and verified, the open items register as at that date with owners, the support window and its expiry, and a statement that the company holds all of its own data in its own systems.
That page takes fifteen minutes to produce and it settles every question anyone will ask about the handover a year later. It is also the clearest possible statement of what the engagement was for. A finance function that cannot be handed over was not being run for the company. It was being run for whoever held the keys.