Utilization rate: the metric that predicts agency failure

Every agency that failed slowly failed the same way: utilisation drifted down for eighteen months and nobody was tracking it, because revenue looked fine and people looked busy.

Utilisation is the share of a person's available time that goes on billable client work. It sounds like an HR metric. It behaves like a financial one, because your costs are fixed and utilisation is what converts them into revenue.

Utilisation = billable hours / available hours
Available hours = contracted hours − leave − public holidays − sick days

Note the denominator. If you divide by 2,080 you'll get a flattering number that nobody can act on, because nobody can work 2,080 hours. Use available hours — around 1,776 for a full-time person with 25 days of leave, 8 public holidays and 5 sick days.

Why it moves profit faster than pricing

Your delivery costs are almost entirely fixed. Salaries, benefits, equipment and your share of overhead don't change if someone bills 20 hours this week or 35. So every additional billable hour is close to pure margin, and every lost one is close to pure loss.

Take a person with a fully loaded annual cost of $144,000 and 1,776 available hours, charged out at $150 an hour.

Utilisation	Billable hours	Cost per billable hour	Revenue	Margin	Margin %
55%	976.8	$147.42	$146,520	$2,520	1.7%
65%	1,154.4	$124.74	$173,160	$29,160	16.8%
75%	1,332.0	$108.11	$199,800	$55,800	27.9%
85%	1,509.6	$95.39	$226,440	$82,440	36.4%

Every ten points of utilisation is 177.6 hours, worth $26,640 a year per person at that rate. On an eight-person delivery team that's $213,120 — more than most agencies would net from a full year of hard selling.

Look at the 55% row. That person is charged out at $150 against a real cost of $147.42. They are, in effect, working for free while appearing fully employed. Nothing in a P&L will show you this. It shows up eventually as "we're busy but there's no money," which is the sentence that precedes most agency collapses.

The rate that isn't achievable

There is a ceiling, and pretending otherwise is how you build a plan that can't be delivered.

At 100% utilisation nobody does timesheets, sales, recruitment, training, line management, internal reviews, or the ten-minute recovery between two intense things. Realistically:

• Junior and mid delivery staff can sustain a high rate, because they have little else to do.
• Seniors and leads carry mentoring, technical or creative direction, and pitch work. Their sustainable rate is meaningfully lower, and pushing it up breaks something else.
• Anyone who also sells cannot be held to a delivery number at all.

Setting one target for the whole agency guarantees either that seniors miss it constantly or that juniors coast. Set it by role, and treat the blended average as the number you report.

The four ways it leaks

Bench time between projects. The most visible cause and the least common in practice, because owners notice an idle person immediately.

Slow starts. A project is signed but the client hasn't sent assets, approved the brief, or scheduled the kickoff. The team is allocated and not billing. Nobody flags it because everybody assumes it's about to start.

Unbilled overrun. The hours happen, the work happens, and it doesn't go on an invoice. This is technically full utilisation and zero revenue, which is why "billable hours" must mean hours you actually invoiced, not hours you logged against a client.

Internal work that quietly grows. The website rebuild, the new case studies, the process documentation. All legitimate. All charged to nobody. It expands to fill whatever capacity is available, which is precisely why it must be capped rather than encouraged.

Measure it weekly or don't bother

Utilisation is a trend metric. A single week means nothing — one person on holiday distorts it. A quarter is too late to act on. Weekly, looking at a rolling four-week average, is the right resolution.

What to track:

Weekly rate    = billable hours this week / available hours this week
Rolling 4-week = billable hours last 4 weeks / available hours last 4 weeks

Two rules make the number honest. Log time the same day, because Friday reconstruction always rounds toward billable. And separate "billable" from "billed" — if you can't put an hour on an invoice, it isn't billable, whatever project code it carries.

Then watch the direction, not the level. A team at 68% and rising is in better shape than a team at 74% and falling, because the second one is spending down a backlog it hasn't replaced.

What to do when it's falling

In rough order of speed:

1. Find the slow starts. Look for allocated-but-not-started projects. Chase the client blockers directly; this is usually a week of someone's time recoverable inside a day.
2. Cap internal work. Give it a budget in hours per person per month and treat overruns like any other overrun.
3. Fix the unbilled overruns. Every hour delivered outside scope is a utilisation number that looks fine and a margin number that doesn't.
4. Change the mix before you change the headcount. Redundancies are the right answer far less often than owners think, and they're irreversible.

This week

Take your last four weeks. For each delivery person, count hours you actually invoiced and divide by their available hours in that period. Not logged hours — invoiced. Then do the same for the four weeks before that.

Two numbers, one direction. If the second is lower than the first, you have somewhere between three and six months to act before it shows up in the bank, and knowing that now is the entire advantage.