How many clients per accountant is the right number? Ask five partners and you will get five answers, usually between 20 and 80. They are all arguing about the wrong thing.
Published benchmarks count the books: transactions, categorisation, reconciliations. In a CAS practice, that is only part of the job. Once the books are reasonably clean, the number that sets your ratio is how much reporting and review each client generates. Change that, and the ratio moves.
What the "how many clients per accountant" benchmarks say
Search the phrase and you find a consistent pattern: wide ranges, with the caveat that it depends.
- A Huskey Practice Manager summary of an Envoice piece puts the typical bookkeeper at 10 to 40 clients, with 30 as the average.
- Growthy puts a full-time solo bookkeeper's ceiling at 20 to 25 clients, from about 112 billable hours a month divided by 4 to 6 hours per client.
- A practitioner on an AccountingWeb thread (an old one, and about small sole traders) was comfortable with 40 per bookkeeper, and 80 only at a real push.
- A recent Closing Entries post quotes published figures of 10 to 15 mid-size clients per bookkeeper without automation and 25 to 40 with AI categorisation.
These are useful as sanity checks. They are also almost all about bookkeeping, and mostly about small clients. Very little of it speaks to a CAS team serving owner-managed businesses that expect a monthly pack and a conversation.
The benchmark measures a third of the job
The Closing Entries author timed their own operation and found that bookkeeping and its close neighbours were only 34% of the work. The rest was client communication, document collection, payments and payroll, which they put at 21 hours per client.
That is one firm, so treat it as an illustration, not an industry figure. But it matches what most CAS leaders see. Automate the coding and your capacity improves, though by much less than the sales deck promised, because the hours that remain are the ones that need judgement or a human reply.
If you build capacity plans on bookkeeping hours alone, you will overload people who look under-loaded on paper.
Count hours, not heads
The better planning method is hours per client, by type. Finmap suggests starting from about 130 productive hours a month after breaks, meetings and task switching, then weighting clients: a simple sole proprietor at 0.5 to 1 hour, a VAT-registered company at 6 to 10, a complex foreign-trade business at 12 or more. Their point is that an accountant with 25 complex clients can be more stretched than a colleague with 45 simple ones.
Be wary of your own starting number too. Tidyflow estimates a senior accountant has roughly 140 billable hours a month, not 176, and says most firms overestimate capacity by 20 to 30% by treating every working hour as productive.
So the arithmetic is simple:
- Take realistic hours per person (130 to 140, not 160).
- Estimate hours per client, by tier.
- Divide.
A team that does this honestly will land well below the optimistic numbers online.
What actually sets the ratio: reporting and review load
Here is what the benchmarks leave out. Once the books are reasonably clean, two things drive the hours per client.
Reporting load. How many packs, ad hoc questions and "can you send me the latest numbers" requests does this client generate? A client who asks for margin by job, cash position and overdue invoices every fortnight costs far more than one who reads a quarterly summary, even if the ledgers are identical.
Review load. How much of the month goes on checking work, chasing exceptions and reading ledgers for problems? If review is a calendar task done on every client regardless of risk, it scales in line with client count. If it is exception-based, it does not.
Both are design choices, not facts of nature. That is the useful news. A firm cannot make its clients simpler, but it can change how much they ask the team, and how much the team has to read to find what matters.
A worked example
This is an illustrative model, not data from a real firm. Take a controller with 130 productive hours a month and clients at 13 hours each.
| Where the 13 hours go | Hours |
| --- | --- |
| Production (coding, reconciling, close) | 5 |
| Reporting (packs, ad hoc requests) | 4 |
| Review and exception hunting | 3 |
| Client conversation | 1 |
Automate production by half and you save 2.5 hours. Capacity goes from 10 clients to about 12.
Now target the reporting and review lines instead. Let clients ask their own questions and pull their own numbers, and let something watch for exceptions so the controller reviews flags, not ledgers. If those two lines fall by half, you save 3.5 hours. Capacity goes to about 13 or 14 clients, and the controller's time shifts toward the one line clients value most: the conversation.
The two moves are comparable in size. The difference is that firms rarely target the second one, because it never appears in a bookkeeping benchmark.
How to move your own ratio
Start with measurement, not tooling.
- Time five clients for a month, tagged as production, reporting, review and conversation. Use your own numbers, not ours.
- Tier your clients by hours, and set a fee floor for each tier. If a client needs 13 hours and pays for 8, the ratio is not your problem. The price is.
- List the repeat questions. Cash position, overdue invoices, margin, spend against budget. Anything asked more than twice a quarter should be answerable without a person.
- Switch review from scheduled to exception-based for your lowest-risk clients first.
This is where tooling earns its place. Pastel, for example, lets clients ask questions of their numbers in plain language and flags what needs attention, which targets exactly the reporting and review lines. It cuts the manual, repetitive work of bookkeepers, payroll and controllers by up to 80%, and the same team serves 70% more clients.
For the hiring side of this decision, see our piece on scaling CAS without hiring ahead of revenue.
The short version
There is no honest single answer to how many clients per accountant. Published ranges run from 10 to 80 because they measure different clients, different services and different assumptions.
For a CAS practice, stop asking about the books. Measure reporting and review hours per client, price to them, and redesign them. That is where the ratio moves.