KPIs every accountancy firm should track

Most firm dashboards measure the past

Ask a firm owner how the practice is doing and you will usually get three numbers: fees, headcount and what is in the bank. All three are real, and all three describe something that has already finished happening. Fees tell you what you billed for work that was done months ago. The bank balance tells you which of those invoices have been paid. Headcount tells you what you decided last year. None of them tells you that four jobs are quietly stuck behind one reviewer, that onboarding has drifted to a month, or that this quarter’s margin has already been decided by write-offs that nobody has recorded yet.

That is the difference between a lagging measure and a leading one. A lagging measure reports the outcome; a leading measure reports the cause, early enough to act on. Profit is the most lagging number in the firm — by the time it moves, every decision that produced it is six months old. Days to onboard, jobs past their target cycle time and the size of the review queue are leading measures: they move first, they move fast, and they are the reason the profit number does what it does. A firm that watches only the outcome will always be explaining its results rather than choosing them, which is the mechanism described in why busy firms aren’t always profitable.

The opposite failure is just as common and considerably more expensive. A firm buys a reporting tool, discovers it can produce ninety measures, and builds a dashboard nobody reads past the second screen. Measurement is not free: every number has to be defined, sourced, checked and argued about, and a number nobody trusts is worse than no number at all because it makes the whole page easy to dismiss. Twelve is about the limit of what a firm of five to fifty people can genuinely keep honest, and twelve is more than enough. If a measure would not change a decision, it does not belong on the page — it belongs in a report you run when the question comes up.

One more rule before the list. A key performance indicator is not a target until somebody owns it. “Lock-up is 95 days” is a fact; “lock-up is 95 days, Sarah owns it, we have agreed 75 by December, and we look at it on the first Monday of the month” is a management system. Numbers without owners produce discussion. Numbers with owners produce change.

The twelve numbers that actually tell you how a firm is running

These are grouped so that no group is allowed to be forgotten. Firms with an accounting background over-weight money and under-weight work; firms in growth mode over-weight clients and miss people entirely. Four groups of three keeps the page balanced.

Three of the twelve are skipped by almost every firm that builds its own dashboard, and they are the three worth insisting on.

• Lock-up days (number 1) is the one number that turns operations into cash. Most firms watch debtors and ignore unbilled work, which is where the larger half usually sits. If a job is finished on the tenth and billed on the twenty-eighth, eighteen days of the firm’s money funded a client for no reason at all.
• The review queue (number 6) is a single count that exposes the most common structural fault in a growing practice: everything funnels through one or two senior people. It is the number that tells an owner whether the firm needs a process change or the management layer described in how to build a management layer in your firm.
• Fee movement at renewal (number 9) is where scope creep becomes visible. A client whose payroll doubled and whose fee did not is not a pricing failure at the point of sale; it is a review that never happened. That is the discipline set out in how to price advisory services.

Two familiar measures are deliberately absent. Utilisation as a headline number rewards people for being busy rather than for finishing work, and in a fixed-fee firm it measures the wrong thing entirely — capacity used against plan (number 10) answers the useful part of the question without the perverse incentive. Net promoter score is absent because at the volumes a typical firm surveys, the score moves on a handful of responses and tells you less than the reasons written next to client losses in number 8.

Worked example: what two of these numbers are worth

The figures below describe an invented ten-person firm with £900,000 of annual recurring fees — an illustration, not a client of ours, and not benchmark data. The arithmetic is the point, because it is arithmetic any firm can run on its own numbers this week.

The lock-up line is straightforward: £900,000 × 95 ÷ 365 is £234,247, and at 60 days it is £147,945. The difference is £86,302 of the firm’s own money currently funding clients rather than sitting in the firm. It is released once rather than earned every year, but it is released without winning a single new client, and for a firm carrying an overdraft it is the cheapest money available to it.

The write-off line is the recurring one. Three percentage points of £900,000 is £27,000 a year of work already done, already delivered and never invoiced — roughly the cost of a junior, given away annually in fragments too small for anyone to notice individually. Neither number appears anywhere in the profit and loss account. The P&L faithfully reports the result of both and explains neither, which is precisely why a firm can read its accounts every month and still be surprised.

The third line does not convert to a clean figure, and it would be dishonest to pretend otherwise. Fourteen fewer overdue jobs is fourteen fewer sources of client chasing, internal escalation and evening work. It shows up later as lower absence, better retention and a January that does not consume the firm — real value that arrives too diffusely to invoice. That is the general case for measuring the operation rather than only the outcome, made at more length in what operating discipline actually means.

Turning twelve numbers into one page people actually use

A dashboard fails for one of three reasons: the data is not trusted, nobody owns the numbers, or looking at it is not part of anything. All three are fixable in a fortnight.

Fix the definitions before the design. Most dashboard arguments are definition arguments in disguise. When does a job start — when the client signs, when the records arrive, or when someone opens it? Does a client who moved from three services to one count as a loss? Write one sentence per measure, agree it once, and never quietly change it, because a redefined measure destroys the history that made it useful.

Take the numbers from the system that already holds them. Lock-up and margin come from the ledgers; cycle time, the review queue and days to onboard come from your practice-management system, provided jobs are actually run in it rather than in inboxes. If the data is not there, that is not a reporting problem to solve with a spreadsheet — it is the requirement that belongs in how to choose practice-management software. In the meantime, a hand-counted number that people believe beats an automated one they argue with.

Give each group one owner. Money, work, clients, people — four names, not twelve. The owner does not have to fix the number alone; they have to know why it moved and what is being done about it. In a firm with no management layer yet, three of the four will be the owner, and seeing that written down is itself a useful piece of information.

Attach the page to a meeting that already exists. Four of the twelve get sixty seconds in the weekly team meeting; all twelve get twenty minutes in the monthly operating review, where each number is either fine, or has an action and an owner against it. Nothing else is required. A dashboard reviewed in a standing meeting survives; one that lives in a folder and depends on somebody remembering to open it does not.

Set targets from your own baseline. Measure for three months before setting a single target, then aim to move each number by a quarter rather than to some external ideal. Published benchmark figures for firm performance vary widely depending on who was surveyed and how the terms were defined, and a target borrowed from one is easy to dismiss the moment it becomes uncomfortable. Your own trend line is harder to argue with and more useful to manage against.

What to do this week

• Calculate lock-up days once, by hand: add unbilled work in progress to debtors, divide by annual fees, multiply by 365. It takes twenty minutes and it is usually the most surprising number in the exercise.
• Count the review queue right now, and note how long the oldest job has been waiting. If nobody can produce that count in five minutes, the count itself is the finding.
• Write down the last twelve months of client losses with a one-line reason against each. Reasons written at a distance are more honest than reasons written on the day.
• Pick the four weekly numbers — jobs past target, review queue, capacity used and cash in — and put them on the agenda of next week’s team meeting, before you build anything.
• Write your one-sentence definition of each of the twelve, and name an owner for each of the four groups. That page is the dashboard; the software is only how it gets filled in.

Firms that do this well are not the ones with the best reporting tools. They are the ones where twelve honest numbers get looked at by the same people at the same time every month, and where each number has somebody’s name against it. The measurement takes an hour a month; the decisions it prompts are what actually change the firm — the same reason a written route beats a good memory in how to build SOPs in an accountancy practice. Setting up that operating rhythm, then running it month after month so it does not quietly lapse in February, is the work described in the Optivo Monthly COO and sequenced in the Optivo method.