When to Move From a Bookkeeper to an Accountant

Why this matters

Owners usually get this decision wrong in one of two directions, and both are expensive. The first is asking a bookkeeper to make calls that are not theirs to make - how to treat a large purchase, how to record a prepayment, whether a worker arrangement is what the shop thinks it is - and then discovering at year end that a full year of decisions rests on a guess. The second is putting a certified accountant on retainer to do work a competent bookkeeper does better and faster, then never asking them anything that uses what they know.

The trigger that actually matters is not revenue. It is not headcount either. It is the kind of question your books are generating, and that can change in a quarter while your size does not move at all. This card gives you one rule and runs it against two shops that come out on opposite sides, in the direction most owners would predict backwards.

Three different jobs, often confused

Bookkeeper. Records what happened, accurately and on a calendar. Reconciles accounts, categorizes transactions, produces the statements, chases the missing documents. A good one is fast, consistent, and asks rather than assumes. This is the job most shops need most of the time, and the quality range within it is enormous.

Controller-level review. Owns whether the close is trustworthy. Designs the accounts, sets the controls, reviews the bookkeeper's work, catches the drift that a person doing the same task every month stops seeing. In a small shop this is almost never a hire; it is a few hours a quarter from somebody senior, either the accountant or an experienced independent.

Accountant. Answers treatment questions, prepares the returns, and advises on anything that changes the shape of the business for more than a year: entity structure, financing, buying or selling, bringing in a partner, adding a line with different economics. In the United States a CPA carries a state license and can perform certain attest work; an enrolled agent is federally credentialed to represent you before the IRS. Which credential you need is a question worth asking directly, and it depends on what you are actually engaging them for.

The confusion that causes damage is treating the third job as a bigger version of the first. It is not a bigger version. It is a different question type, and no amount of bookkeeping skill converts into it.

The gate, and what it deliberately ignores

Applied per shop, over a trailing 12 months, and either half is sufficient on its own:

(a) You have had 3 or more treatment questions - questions where the right answer is set by rule rather than by preference, and where the answer changes a number you have already reported; or

(b) You have had any single structural event that changes the entity's obligations, ownership or financing for more than a year.

Three is the threshold rather than one because a single treatment question a year is normal and gets handled inside the annual filing relationship at no extra cost. Three in a year means they are being generated by something structural in how the shop now operates, and the fourth one is already forming while you are reading this.

The gate ignores revenue, headcount, truck count and years in business, deliberately. Those correlate with question volume loosely and with question type barely at all. A shop can triple in size doing exactly one thing in exactly one jurisdiction and never generate a treatment question. A shop can stay two crews and generate five in a year by adding one new revenue model.

It also ignores whether the books are messy. Messy books are a bookkeeping problem, and hiring an accountant to fix them is paying a specialist rate for clerical work that will re-mess itself the following month because the underlying process never changed.

Case one: two crews, and it resolves yes

A residential service shop, 2 crews, roughly flat headcount for two years. In the trailing 12 months:

• Launched prepaid annual maintenance agreements, which raised how prepayments should be recorded. Treatment question 1.
• Financed two vehicles, which raised how the payments split and how the assets appear. Treatment question 2.
• Began taking work in a second jurisdiction, which raised where and how sales tax applies to their service mix. Treatment question 3.
• Started paying one long-time helper as a subcontractor rather than an employee, which raised whether that arrangement is what they believe it is. Treatment question 4.
• Wrote off a receivable large enough to move a month, which raised the timing of when it is treated as uncollectible. Treatment question 5.

Five treatment questions against a threshold of 3, so half (a) is met with room. The vehicle financing is also a multi-year obligation, so half (b) is independently met as well. This shop clears the gate on both halves, not just one, and it is the smaller of the two shops in this card.

What made the difference was not growth. It was that four separate decisions in one year each introduced a new kind of transaction. Every one of them was reasonable on its own. Together they moved the shop from one business model to four overlapping ones, and the books stopped being a record of a thing anyone had already decided how to record.

Case two: six technicians, and it resolves no

A second shop, 6 technicians and an office admin, roughly three times the first shop's revenue, running the same single trade in one jurisdiction for four years. No financing, no agreements, no subcontractors, no new lines. In the trailing 12 months:

• Bought a piece of equipment large enough to raise a repair-versus-asset question. Treatment question 1.

One question against a threshold of 3, so half (a) fails. No structural event, so half (b) fails. Both halves fail and the gate resolves no.

What this shop actually had was a different problem wearing the same clothes. The close ran 5 to 6 weeks late every month, the bank reconciliations carried small unexplained differences that got plugged, and the owner had stopped reading the statements because by the time they arrived they described a month he could no longer do anything about. That reads like "we need a real accountant" and it is not. It is a bookkeeping capacity and process problem, and putting an accountant on retainer would have produced expensive, correct, still-late statements.

The single treatment question got sent to the preparer who already handles the annual return, answered in one exchange, and cost nothing extra. That is the correct handling of a below-threshold question and it is why the threshold is 3 rather than 1.

What each shop did next

The two-crew shop engaged its accountant on a defined cadence rather than a general retainer: one hour a quarter scheduled in advance, plus named access for treatment questions as they arise, plus the annual return. The quarterly hour has a fixed agenda - the five open treatment questions, the balance sheet, and anything the bookkeeper flagged - which is what keeps it an hour. They did not replace the bookkeeper, and that is the part owners most often get wrong. The bookkeeper still does the recording; the accountant now decides how a new kind of transaction gets recorded, once, and then the bookkeeper repeats it correctly forever.

The six-technician shop kept its annual filing relationship and spent the money on the actual constraint: it moved the close onto a fixed calendar with a document packet due on a set business day, and it bought roughly 3 hours a quarter of controller-level review to check the reconciliations and the account structure. The close came in under two weeks within a quarter. The owner started reading the statements again because they now arrive while the month is still actionable.

Both shops spent more than they had been spending. They spent it on different things, and if you swapped the two solutions both would have been worse off: the small shop would have had a beautifully run close producing correct-looking numbers built on five unresolved treatments, and the larger shop would have had an accountant on call with nothing to ask them.

The signal you over-hired

Two symptoms, and they show up within about six months.

The first is that your accountant's work product is almost entirely descriptive. If what comes back each quarter is a restatement of what your own statements already say, with no decision named and no question raised, you are paying advisory rates for reporting you already have. Redirect the engagement or shrink it. A useful quarterly hour ends with at least one thing you will do differently.

The second is that treatment questions have stopped arriving. If the shop has been stable for a year, the model has not changed, and the last three quarterly meetings had no agenda beyond reviewing the numbers, the shop has moved back below the gate. That is not a failure, it is a normal cycle - businesses generate treatment questions in bursts around change and then go quiet - and the right response is to drop to annual filing plus named access and re-run the gate when something changes.

The mirror mistake is worse and harder to see: a shop that clears the gate, engages nobody, and continues to have its bookkeeper answer treatment questions by picking whatever seems reasonable. There is no alarm for that. The books look finished. The finding arrives a year or two later, in a lender review or a due diligence request, and by then the wrong treatment is in every month you have reported.

References

• IRS, representation rights of enrolled agents and other credentialed preparers
• See related: The First Office Hire: CSR vs Dispatcher vs Bookkeeper Decision Matrix, How to Ask Your Accountant a Question Worth Their Time, The Month-End Document Handoff SOP, How to Change Bookkeepers Without Losing History