How to build a management layer in your accountancy firm

What a management layer actually is

A management layer is not an org chart, a title, or a pay rise for your longest-serving senior. It is a set of decisions that used to reach the owner and now stop one level below. That definition is worth holding on to, because firms that miss it end up with people called managers who still queue outside the owner’s door — all of the salary cost, none of the relief.

In a founder-led accountancy firm the decisions that clog the owner’s week are strikingly consistent. What do we quote for a job that is slightly non-standard? Is this set of accounts good enough to go out? What do we do about the client who still hasn’t sent their records? Can a junior take the week off in January? Which of these six urgent things happens first? None of them is strategic. Every one of them is urgent to somebody. Together they are the reason the owner never gets to the work only the owner can do.

Building the layer means naming those decisions, deciding who owns each one, giving that person the authority and the information to make it, and then — the genuinely hard part — living with the answer they give. Recruitment, if it is needed at all, comes after that, not before.

The test: count decisions, not headcount

Before you write a job advert, run a one-week tally. Every time a question reaches you — in person, on Teams, by email, on the way to the kettle — write one line: who asked, what about, and roughly how long it took to deal with. Do not change your behaviour while you are measuring. The point is to see the firm you actually have, not the one you would describe to a stranger.

At the end of the week, sort every line into three buckets:

• Genuinely yours. Pricing policy, partner-level client relationships, strategy, hiring, anything that shapes the firm rather than runs it.
• Someone else’s, blocked by authority. They knew the answer perfectly well. What they needed was permission.
• Someone else’s, blocked by skill or information. They could not have answered it, or could not see the numbers, the file or the deadline they would have needed.

The proportions tell you what to buy, and the answer is often not a person. If the second bucket dominates, you have an authority problem rather than a headcount problem, and it is by far the cheapest thing on this page to fix: a written list of decision rights and a fortnight of the owner refusing to answer questions that are no longer theirs. If the third bucket dominates, you have a capability or systems gap, and the honest fix is training and documented process — start with how to build SOPs in an accountancy practice and the trapped capacity described in increasing capacity without hiring. Only when the first bucket is small, and the other two are large and spread across several people at once, is a management layer the right purchase.

Promote or recruit?

Assume the tally says you genuinely need a manager. The next decision is where they come from, and it is not a close call in most firms — but it is worth making deliberately rather than by default.

The most common mistake is promoting the best technician and then changing nothing else about their week. If a new manager keeps a full portfolio and a full chargeable target, management happens after six o’clock, which means it does not happen. Decide in advance how much of their fee-earning target comes off, write it down, and tell the rest of the team what it has been replaced with.

Worked example: what a manager really costs

Firms consistently underestimate this because they think in salary. The example below is illustrative — one manager on £55,000 — but the tax and pension figures are the real ones for the 2026/27 tax year, which runs from 6 April 2026.

Two points about that total. First, the rates are current, not rules of thumb: employer’s secondary Class 1 National Insurance is 15% and the secondary threshold is £5,000 a year for 2026/27, and the auto-enrolment minimum employer contribution is 3% of qualifying earnings between £6,240 and £50,270 for the same year. Second, the Employment Allowance — £10,500 for 2026/27 — reduces an eligible employer’s National Insurance bill for the year as a whole, not employee by employee. A firm with an established payroll has almost always absorbed the whole allowance against existing staff long before this hire arrives, so that £7,500 is real cash leaving the business.

Then add the lines the salary never shows: recruitment, whether that is an agency fee or forty hours of your own time; a desk, a laptop and licences; and the cost of the seat sitting empty while you look. The largest hidden line is your own attention. Assume three hours a week of owner time for the first thirteen weeks of induction, coaching and correction — thirty-nine hours. At an illustrative £150 an hour for the owner’s time, that is £5,850 invested before the layer returns anything at all.

Now the return side, on the same illustrative basis. Statutory paid holiday is 5.6 weeks a year — 28 days for someone working five days a week — so a working year is around 46 weeks; call it 44 once training and sickness are allowed for. If the manager permanently removes six hours a week of decisions, review and chasing from the owner, that is 264 hours a year, or £39,600 of owner time at the same rate.

Notice that £39,600 does not by itself cover £63,821. That is the honest arithmetic, and it matters: a management hire is never justified by the owner’s freed time alone. It is justified by that freed time plus the manager’s own output — the jobs they run, the reviews they clear, the juniors who stop waiting — and it only works if the recovered owner hours are genuinely redirected into advisory, better clients or growth. If those hours are absorbed back into being busy, the hire is a cost with no matching return, which is exactly how firms end up with a bigger payroll and the same bottleneck. The capacity argument is set out in more detail in how to make your firm less founder-dependent.

Write the decision rights down before the job description

A job description describes activity. A decision-rights list describes authority, and authority is the thing that is actually missing. One page is enough. For each decision, state who decides, what the limit is, and who simply needs to be told afterwards.

The thresholds in that table are examples, not recommendations — the number matters far less than the fact that a number exists and everyone knows it. What kills a management layer is not a limit set slightly too low. It is a limit that was never set, so every judgement call defaults back to the owner.

The first 90 days

Install the layer in a sequence, not all at once. A workable rhythm:

• Weeks 1–2 — publish. Circulate the decision-rights page to the whole team, not just the manager. A layer the team does not know about does not exist, and people will keep walking past it to the owner.
• Weeks 3–6 — redirect. When a question arrives that now belongs to the manager, the owner’s answer is one sentence: “That’s yours — what do you think?” Then wait. This is the single hardest habit in the whole exercise and the one that decides whether it works.
• Weeks 3–12 — a weekly one-to-one. Thirty minutes, same slot, never cancelled: decisions made, decisions stuck, what the manager needs. Cancelling it twice tells the team the layer is optional.
• Weeks 7–12 — back the calls. The manager will make a decision the owner would have made differently. Unless it is genuinely damaging, support it in public and discuss it in private. Overruling a manager in front of the team once undoes a quarter of work.
• Day 90 — re-run the tally. Repeat the one-week decision count. If the second bucket has shrunk, the layer is real. If it has not, the authority was announced but never actually transferred.

Why management layers fail

Four failures account for most of it, and none is about the calibre of the person appointed.

• Nothing came off their plate. A full portfolio plus a management role equals an overworked senior with a new title.
• Authority granted, then quietly withdrawn. The owner keeps answering questions that were delegated, usually because answering is faster than coaching. It is faster this week and slower every week after.
• No rhythm. Without a fixed weekly conversation, problems surface only when they are large, which pulls the owner back in and confirms everyone’s suspicion that the layer never worked.
• Undefined success. If nobody agreed what the manager is accountable for — jobs out of review by a date, deadlines met, juniors not waiting — then any outcome can be argued about, and usually is.

What to do this week

• Run the one-week decision tally. One line per question. No behaviour change while measuring.
• Sort the lines into the three buckets and count each one.
• Write the decision-rights page for whichever bucket is largest — even if you never hire anyone.
• Put a recurring thirty-minute one-to-one in the diary with whoever is closest to being that manager today.
• Cost the hire honestly using the figures above before you commit, and be clear what the freed owner hours will be spent on.

Most firms find the tally alone changes how the next quarter runs, because it makes an invisible problem countable. Doing the rest of it — naming the decisions, moving them, and holding the line while people test whether you meant it — is steady operating work that rarely survives a busy season without someone whose job it is. That is precisely the work the Optivo Monthly COO exists to do, and how it is sequenced is set out in the Optivo method.