A founder I know walked into a board meeting last spring with the best utilisation number the firm had ever posted. Ninety-one percent across the practices, up from the high seventies a year earlier. The board liked it. Same quarter, margin softened by four points, two delivery teams were quietly burning out, and the head of delivery had started declining work the firm should have wanted. The number going up and the business getting worse were not a coincidence. The first was partly causing the second.
That is the uncomfortable thing about bench utilisation. It is the metric almost every services CEO watches as the headline measure of health, and it is the one most likely to be lying to them. High utilisation feels like the business running hot, so leaders push it higher. But utilisation does not measure whether the work is priced right, whether you can take the next deal, or whether delivery is about to crack. It measures how full the calendar is. The CEO is reading a number about activity and treating it as a number about health.
The utilisation metric is a proxy that drifted
Utilisation earned its place honestly. For a services firm, billable time is the product, and the share of available time you can bill is a fair first read on whether the machine is working. Low utilisation usually did mean trouble: idle people, thin pipeline, money walking out the door on salaries the work was not covering. So leaders learned to watch the number, and watching it was reasonable.
The trouble is what the number was ever standing in for. Utilisation was never the thing anyone actually cared about. It was a proxy for three questions that are harder to see directly. Are we making money. Can we take on more work. Is delivery about to break. For a long time a single utilisation figure tracked all three closely enough that you could read one and infer the other three. That is why it became the headline.
Proxies drift. The business got more complex, the deals got more mixed, the practices multiplied, and the tight relationship between "how full are people" and "is the firm healthy" loosened. The number kept being reported because it was easy to compute and everyone already trusted it. But it slowly stopped tracking the three questions it was hired to answer. Now you can read 90% and be wrong about all three. The metric did not break. It just quietly stopped meaning what the board thinks it means.
A high utilisation number says nothing about price
Start with margin, because this is the one that stings most.
Utilisation tells you your people are busy. It tells you nothing about whether the work they are busy on makes money. A consultant booked solid on a deal that was discounted to win, priced at a blended rate that never covered the senior time it actually consumed, reads as 100% utilised. The dashboard glows. The deal loses money on every hour billed.
Two firms can post identical utilisation and run completely different P&Ls. One fills its calendar with well-priced work where the rate holds and the grade mix matches what was sold. The other fills the same calendar with under-priced work it took because the bench was making the founder nervous. Same number. Opposite health. Utilisation cannot tell them apart, because it counts hours, not the margin on those hours.
This is why pushing utilisation up can actively destroy margin. When the headline metric rewards filling time, the rational move when the bench grows is to take whatever work clears the room. Discount to close. Say yes to the deal that needs senior people you priced as mid. The utilisation number rewards exactly the behaviour that bleeds the firm, and it does it while looking like good management. You optimised the proxy and broke the thing it was supposed to protect.
Ninety percent utilisation means zero headroom for the pipeline
Now capacity, which is where the high number is most actively dangerous.
A services firm does not deliver only the work it has signed. It is always carrying a pipeline it intends to win, and that pipeline is going to land on the same people the utilisation report says are already full. Utilisation measures the firm against signed work. It is structurally blind to demand that has not closed yet.
So 90% utilisation does not mean "healthy and efficient". It means "almost no room for the next three deals in the pipeline". The fuller the calendar, the less headroom the firm has to absorb the work it is actively trying to sell. A firm running hot on utilisation and proud of it can be one signed deal away from a capacity crisis, because the metric it is celebrating is the absence of the slack it needs to take on what it is selling. Practice leads feel this before the CEO does. Their recurring complaint, in firm after firm, is visibility of pipeline: what is coming at us, and how many people will we need. Utilisation answers neither.
This is the trap of treating a single fullness number as the goal. Push it to 100% and you have not maximised efficiency, you have eliminated the buffer that lets you say yes to growth. The number that looks like peak health is the number that means you cannot take the next deal.
Slack is not waste, and the metric calls it waste
The same blindness shows up one layer down, in how utilisation treats the gaps.
Every realistic delivery plan carries slack. Time held back because a committed project might expand, because the senior architect cannot be 100% booked or the first slip takes the whole quarter down, because one delayed project creates downstream capacity conflict and someone has to absorb it. That slack is not idle waste. It is the insurance that lets the firm survive its own delivery risk.
Utilisation reads all of it as a problem. An unbooked hour is an unbooked hour, whether it was genuinely wasted or deliberately protected. A firm optimising the number will drive that slack out, because every hour of it drags the headline down. Then the first project slips, there is no buffer to absorb it, and the slip cascades through every adjacent deal sharing those people. The metric punished the one thing standing between a normal delay and a bad quarter.
This is the part that makes utilisation worse than merely uninformative. It is not neutral. Optimising it directly causes some of the harms it is supposed to warn you about. It rewards filling time over protecting margin. It rewards eliminating the headroom you need for pipeline. It rewards burning the slack that absorbs delivery risk. A leader who trusts the number and pushes it is not just missing the real picture. They are actively steering toward the crisis the number was meant to prevent.
What to watch instead of how full the calendar is
If utilisation answers the wrong question, the fix is not a better single number. It is watching the three questions directly instead of through a proxy that no longer tracks them.
For margin, stop reading utilisation as a profit signal and read margin where it actually lives: by deal and by practice. A firm can be busy and unprofitable, and the only way to see it is to look at what each deal and each practice is actually returning, not at how full the people on them are. The question is not "are they busy". It is "is the work they are busy on making money".
For capacity, watch demand against signed-plus-pipeline, not just signed. The useful view weights the open pipeline by probability and lays it on top of committed work, role by role, so you can see where the next two quarters get tight before the deals close. Headroom, the gap between what you can deliver and what you have already committed, is the number that tells you whether you can take the next deal. Utilisation cannot show you headroom, because it has no concept of the demand that has not signed.
For delivery risk, treat protected slack as a planned position, not a failure. The question is not "is everyone booked" but "do we have the buffer to absorb the slip we know is coming". A plan with deliberate headroom on the rare roles is healthier than one at 100% across the board, even though utilisation scores the second one higher.
None of these is a single headline figure, which is exactly why utilisation won in the first place. One number is easier to report than three honest views. But the easy number is the one that drifted, and the firms that keep getting surprised are the ones still managing the proxy because it fits on one slide.
Where the real questions become visible
Watching margin, headroom, and delivery risk directly is a discipline before it is a tool. You can run it off honest spreadsheets if someone maintains them, the same way you can run capacity planning off a sheet until the practices multiply and the reconciliation eats a senior person's Friday.
This is the part of Estii built for the real questions rather than the proxy. The Forecasts view reads capacity against signed-plus-pipeline at three levels, Confirmed for committed work, Forecast for the probability-weighted pipeline, and Exposure for the if-everything-closes ceiling, so headroom is visible against the pipeline you are actually selling into, not just the work already signed. Because it reads from the same commercial model sales builds the deal in, margin sits on the pipeline by deal and status next to value, so "busy" and "profitable" stop being the same number. The point is not a better utilisation gauge. It is that the questions utilisation was always standing in for become things you can look at directly.
Bench utilisation will keep being reported, because it is easy and the board already trusts it. That is fine, as a footnote. The mistake is treating the fullness of the calendar as the health of the firm. A firm can be at 90% and under-priced, over-committed, hiring blind, and one slipped project from a capacity crisis, with every one of those facts invisible in the number on the slide. The metric measures activity. Health is a different thing, and it has always lived in the questions the proxy quietly stopped answering.
