What is a utilisation target actually measuring?
Utilisation is chargeable hours divided by available hours. It was designed as a crude test of one thing: is the capacity a firm pays for being pointed at work a client will pay for. That test made sense in a world where the unbillable part of the week was treated as a fixed and largely unmanageable overhead. Everybody had write-ups to do, files to note, reports to assemble and timesheets to piece together on a Friday. Nobody expected that block to move, so the sensible question was how much of the remaining time each person sold.
Two assumptions sit underneath that. The first is that hours are a reasonable proxy for value delivered. The second is that the denominator behaves, in other words that the non-chargeable share is roughly stable across people and across years. Rebuilding a process attacks the second assumption directly, and once it goes, the number stops carrying the meaning partners have spent careers reading into it.
What happens to the number when the admin disappears?
Take a fee earner who loses several hours a week of writing up, chasing and formatting. On hourly billing, nothing about their output has to change for utilisation to climb. The chargeable numerator stays where it was and the unbillable drag falls away, so the percentage rises on its own. A partner reading the monthly pack sees a team apparently working harder. Nothing of the sort happened. The measure simply stopped counting work it used to count.
Now take the opposite case, which is more common than firms expect. The engagement is a fixed fee. The report that took a day to assemble now takes an afternoon. Chargeable hours per job fall, so utilisation falls, while the margin on that job improves. The firm has just become more profitable and its headline efficiency measure has gone down. If the target is enforced without thought, someone will be asked to explain a good result as though it were a failure.
Both movements are artefacts of the definition. Neither tells you whether the firm delivered more, delivered faster or earned more per engagement.
Why does realisation tell you more than utilisation?
Realisation, the share of recorded time that survives to the invoice, has always been the more honest of the pair, because it includes the client's judgement. Time that gets written off was time the client would not pay for, whatever the timesheet said. When admin shrinks, realisation behaves in a more interesting way than utilisation does.
Some of what is written off is padding and guesswork: hours logged at the end of the week from memory, rounded up, then trimmed by a partner who knows the matter will not bear them. Capture at the point of work removes a chunk of that, which usually means recorded time gets closer to reality on both sides. See timesheet capture for what that looks like in practice.
The more useful number of all is effective rate: fee divided by the hours actually spent, by engagement and by work type. It survives the shift from hourly to fixed fee, it is comparable across a team, and it moves only when something real happens. We take the pricing side of that argument further in pricing work that AI makes faster.
What goes wrong if you leave the target unchanged?
Targets are instructions, whatever the memo says. Leave a chargeable hours target in place while removing the unbillable work that used to justify it, and three things tend to follow.
- Hours get found. Work that once took half a day expands to fill the slot, because the slot is what gets measured. The firm pays for the efficiency and never collects it.
- Recording drifts. If nobody trusts the numerator, the numerator stops being trustworthy. That damages pricing for years, because next year's fee estimate is built on this year's recorded time.
- The wrong people are praised. Whoever has the most hourly billed work looks best, regardless of what the firm earned per engagement.
There is a subtler cost. Senior capacity freed by removing write-ups is the scarcest thing most firms have, and it is the thing a utilisation target is least able to direct. That capacity is exactly what clears the queue described in the partner bottleneck, and a percentage on a dashboard will not send it there.
What should sit alongside utilisation?
Not a replacement, a fuller set. Four measures do most of the work for a firm of 20 to 200 people.
- Realised fee per fee earner. The blunt commercial answer. It ignores how the hours were coded and asks what the person's capacity actually earned.
- Effective rate by work type. Fee divided by hours spent, split by the kinds of work you repeat: reports, bids, reviews, onboarding. This is where the effect of a rebuilt process shows up first.
- Turnaround. Days from the trigger event to the deliverable leaving the building. Clients feel this long before they feel anything on your P&L, and it improves most visibly when client reporting or write-ups stop queuing.
- Allocated capacity. Hours deliberately committed to named non-chargeable purposes: pitching, supervision, knowledge work. Making these an intention rather than a residue is the biggest change of mindset here.
The capacity calculator gives a rough view of what a change in the unbillable share does to headroom across a team, and measuring return on AI sets out how to tie that back to money without inventing a benefit.
How do you rebaseline without losing your comparisons?
Do the boring thing first. Before anything is rebuilt, record the current unbillable share by role for a normal month, using the method in how to calculate unbillable hours. That is your reference point, and without it every later argument becomes anecdote.
Then run both definitions in parallel for a quarter after go-live: the old one, with the removed admin notionally added back, and the new one. The gap between them is the effect of the change, stated plainly, and it stops anyone claiming credit for a movement the definition created.
Finally, decide in advance what the freed hours are for, and write it down. More matters at the same headcount. Faster turnaround as a service promise. Bid work the firm has been declining. Supervision time that juniors have been missing, which matters more than it used to, as we argue in what changes for junior staff. A target that reflects an intention is worth keeping. A target that survives only because it has always been there will quietly undo the work.
Where does this bite hardest?
Hardest in firms where the measure is culturally load bearing. In law firms, chargeable hours sit inside appraisal, progression and pay, so changing the denominator without changing the conversation is genuinely difficult. In accountancy firms, much of the work is already fixed fee, so effective rate per job is the natural measure and utilisation is closer to a staffing check than a performance one.
Either way the sequence is the same: measure the unbillable share honestly, rebuild the process that carries most of it, rebaseline the target, then manage the freed capacity on purpose. The unbillable hours calculator gives you the starting figure, and the audit will tell you which process is holding the hours.