Year-end season capacity planning for small practices

• Capacity planning is counting: jobs times hours, laid against the months they actually land in.
• The January self assessment peak is the problem everyone knows about and almost nobody plans for in hours.
• Practices recruit in June for a January they will have forgotten by April. That is how you end up overstaffed in May.
• Your four options are overtime, a temp, a per-job outsourcer, or a standing seat you flex up. Each has a cost and a failure mode.
• The capacity calculator does the arithmetic in about two minutes.

Start by counting, not guessing

Most small practices plan year-end season by feel. The partner knows it is going to be bad, tells everyone to brace, and then it is bad. Nobody has written down how many hours of work are coming or when.

The fix is dull. List every recurring job. Put an hours figure next to each one. Put a month next to each one. Add it up by month. That is your demand curve. It takes an afternoon, and once you have it you can argue about the numbers instead of arguing about how tired everyone feels.

If you want a shortcut, the capacity calculator takes your client counts and typical hours per job and gives you the monthly total. Use your own hours per job if you have them. If you do not, start with the defaults and adjust once you have timed a few real jobs.

What to count

• Limited company accounts and CT600s. Count each set. Group them by year-end month, because that decides when the work lands.
• Self assessment returns. Count them separately from accounts, because almost all of them land in one window.
• Sole trader and partnership accounts. These often arrive with the tax return and get done in the same sitting, so count the hours together.
• VAT returns. Quarterly, so they never stop. They are the background noise your peak has to sit on top of.
• Payroll. Weekly and monthly runs that will not move for anybody.
• Bookkeeping. Monthly or quarterly, and the jobs you quietly push back when accounts are due.
• Everything else. Onboarding, chasing records, confirmation statements, queries, calls. Put a number on it even if it is a guess. Ignoring it is how the plan breaks.

Hours per job

Be honest. A small limited company with tidy Xero records might be four to six hours from start to signed-off. One with a carrier bag of receipts might be fifteen. A simple self assessment might be an hour and a half. One with rental income and a partnership share might be four.

Use an average per category, but do not use the best case. Use the typical case. Then add a review allowance for the partner or manager, because that time is real and it is usually the bottleneck.

Map it against the calendar

Now lay the hours against the months. Accounts land after the year-end, with a filing window set by Companies House. Self assessment has a fixed filing window set by HMRC. We are not going to state the deadlines here because they are not ours to state and they change; check them on gov.uk and the Companies House pages and plan from the current dates.

What matters for planning is the shape, not the exact date. Company year-ends cluster, often around March and December. Self assessment returns mostly arrive late, whatever you do, so the back end of the window is the peak. VAT and payroll sit underneath all of it and do not move.

An illustrative year for a three-person practice

The table below is made up to show the shape. It is not a benchmark and it is not a client. Three people, say 400 hours a month of productive capacity between them once you strip out admin, holiday and the partner's selling time. Swap in your own numbers.

Look at the shape. Seven months with spare hours. Four months short, two of them badly. Total demand over the year is roughly 4,460 hours against roughly 4,580 capacity. On paper the practice has enough people. In practice it is drowning in December and January and quiet in February and March.

That is the whole problem with year-end season. It is not a headcount problem. It is a timing problem.

Why practices hire in June for January

Here is the cycle. January is brutal. In February the partner swears never again. In March a job advert goes up. Recruiting takes a while, so the new person starts in June. They spend the summer, the quiet season, learning the job on work that was never short of hands. By the time January arrives they are useful, and January is better.

Then it is April again, the pain has faded, and the practice now has a permanent salary on the books for a peak that lasts ten weeks. Using our working rule of thumb that a UK employee costs about salary times 1.35 all-in, a bookkeeper on £27-32k (Indeed and PayScale salary data, 2026) is around £36-44k a year. You have paid for twelve months to solve four.

Nobody does this on purpose. It happens because the hiring decision is made when the pain is fresh and reviewed when it is not. The counting exercise above stops that, because the numbers do not care what month it is.

Your four options for the peak

1. Overtime

Cheapest on paper. Your existing team knows the clients and the software. The cost is burnout and error. Tired people make mistakes on returns, and the mistakes surface in March when nobody has the energy to fix them. Overtime works for a short, sharp push. It does not work for a peak that lasts a quarter, and it gets worse every year you rely on it.

2. A temp

Agencies will find you a bookkeeper or a part-qualified for three months. You pay an agency margin on top of the day rate, and you get someone who has never seen your clients, your working papers or your software settings. In a good year they are productive by week three. In a bad year they leave in week two for a permanent job. Temps are fine for pure processing if you can supervise them. They are a risk on anything that touches a client.

3. A per-job outsourcer

Send the job, pay per set. Offshore per-job rates on the market run about £120-300 for a small limited company set and £15-40 a self assessment return, which is our read of what providers publish. The attraction is obvious: no commitment, pay for what you use. The catch is that every other practice in the country is sending work in the same window, turnaround stretches, and you are managing a queue you cannot see. Each job also arrives with a stranger's queries. Per-job works when volume is low and the records are clean. It wobbles when you are sending forty jobs in a fortnight. We wrote more about this in the comparison of outsourcing models.

4. A standing seat you flex up

This is the Muckin model, so take it with the bias in mind. A dedicated person, on your work all year, in your software, who knows the clients by October. A part-time Accounts seat at 40 hours a month is £850; a full seat at 160 hours is £2,650. You run part-time through the quiet months and step up to half or full for the peak, with a month's notice either way. There is no lock-in.

Go back to the table. The three-person practice is short by roughly 300 hours across November, December and January and has 400-odd hours spare across the spring. A part-time seat through the year plus a step-up to full for the peak months covers the gap without a twelfth salary. The point is not that a seat is magic. It is that you are buying hours where the hours are needed, from someone who already knows the work. See how it works for the review and sign-off process.

A simple way to decide

1. Count the hours. Use the calculator or a spreadsheet. Either is fine; what matters is that you do it.
2. Find the gap months and the spare months.
3. If the gap is under about 40 hours in any month, overtime will probably do it.
4. If the gap is 40-150 hours for two or three months and your records are clean, per-job outsourcing can work if you start sending early.
5. If the gap is bigger, lasts longer, or you want the same person doing it every year, a standing seat that flexes is cheaper than a hire and calmer than a temp.
6. Re-run the count every summer. Do it in July, when you are calm, not in February, when you are angry.

What usually goes wrong

• Forgetting review time. Production hours are only half the story. If one partner reviews everything, the partner is the ceiling. Count their hours too.
• Assuming records arrive on time. They do not. Plan on the back half of every window being the heavy half.
• Not protecting VAT and payroll. These do not move. If accounts season eats the bookkeeper, VAT returns get rushed, and those errors cost money.
• Starting outsourcing in December. Any outside help, seat or per-job, needs time to learn your way of doing things. Start in September at the latest.
• Only planning January. Corporation tax season is usually the bigger total. January is just the loudest.

Where to start

Run your numbers through the capacity calculator. It will show you the gap by month and what a seat would cost against it. If the shape looks like the table above, get in touch and we will talk through which seat size fits and when to step it up. We muck in from September if you want the peak covered properly.

Published 22 August 2026. Tax rules and rates change — check current figures on gov.uk before relying on anything here. This is general information for practice owners, not advice.