What does 'on the bench' mean at an agency?

A person is on the bench when they are employed and available but have no billable client work assigned. It can be a few hours between tasks or several weeks between projects. The agency pays their full cost during that time and earns nothing from it directly.

How much bench time is healthy?

Most agencies I know plan for delivery staff to bill between 70 and 80 percent of their available hours, which leaves 20 to 30 percent for internal work, training, sales support, and gaps. Below about 65 percent for more than a quarter, the agency is overstaffed for its revenue. Above 85 percent for long, people burn out and there is no room to start new work.

How do you calculate the cost of the bench?

Divide a person's loaded annual cost by their available hours to get a cost per hour, then multiply by the unbilled hours. A developer who costs $120,000 across 1,800 available hours costs $66.67 an hour, so 150 bench hours cost $10,000 in salary. The revenue you did not earn in those hours is a separate, larger figure.

What should employees do while on the bench?

Work from a prepared, prioritized list: support for live sales opportunities, reusable components and internal tools, training in skills you plan to sell, and documentation. Every item should have an owner and a definition of done. Unstructured bench time is costly and demoralizing for the person sitting on it.

Bench management: what idle time costs an agency and what to do with it

Every agency has a bench, including the ones that claim they do not. If your people are employees, there are hours you pay for and cannot bill. The only choice is whether you measure those hours and decide what happens in them.

This article is part of the Team hub. It covers the cost of the bench, the amount that is healthy, what to do with the time, and how to see a gap coming.

What the bench costs

Start with one person. A developer has a loaded cost of $120,000 a year: salary, payroll taxes, benefits, equipment. After holidays, vacation, and sick days, they have about 1,800 available hours.

Line	Calculation	Amount
Loaded cost per year		$120,000
Available hours		1,800
Cost per available hour	$120,000 / 1,800	$66.67
One month on the bench	150 hours
Salary cost of that month	150 x $66.67	$10,000
Revenue not earned at $150 an hour	150 x $150	$22,500

One developer, one month, $10,000 out of pocket and $22,500 of revenue that never happened. Both numbers matter and they answer different questions. The salary cost tells you how fast the bench consumes cash. The lost revenue tells you what a filled month would have been worth.

Now the whole team. Ten delivery people at the same cost:

Line	Calculation	Amount
Total loaded cost	10 x $120,000	$1,200,000
Total available hours	10 x 1,800	18,000
Billed hours at 75% utilization	18,000 x 0.75	13,500
Bench hours	18,000 - 13,500	4,500
Salary cost of the bench	4,500 x $66.67	$300,000
Revenue at $150 an hour	13,500 x $150	$2,025,000
Gross profit	$2,025,000 - $1,200,000	$825,000
Gross margin	$825,000 / $2,025,000	40.7%

A quarter of the delivery payroll, $300,000, pays for hours that produce no invoice. That is normal. It is also the largest discretionary number in the business, and most owners have never written it down.

Here is why it deserves attention. One percentage point of utilization on this team is 180 hours. At $150 an hour, that is $27,000 of revenue a year with no added cost.

Utilization	Billed hours	Revenue	Gross profit	Gross margin
65%	11,700	$1,755,000	$555,000	31.6%
70%	12,600	$1,890,000	$690,000	36.5%
75%	13,500	$2,025,000	$825,000	40.7%
80%	14,400	$2,160,000	$960,000	44.4%

The move from 70 to 80 percent adds $270,000 of gross profit on the same payroll. The drop from 75 to 65 removes the same amount. Few other levers in an agency move profit that far. The metric is explained in full in the Bill hub’s article on utilization rate, and you can put your own figures into the utilization calculator.

How much bench is healthy

Zero is the wrong target. An agency at 100 percent utilization cannot start a new project without delaying an existing one, cannot absorb a sick week, has nobody free to help with a proposal, and is training no one. It also loses people. Sustained utilization above 85 percent for delivery staff has ended in resignations every time I have watched it.

The range I plan to, for people whose main job is delivery:

Utilization over a quarter	What it usually means
Above 85%	Overloaded. Quality and retention at risk. Add capacity or slow sales
75% to 85%	Healthy for most agencies
65% to 75%	Acceptable in a slow quarter. Watch the forecast closely
Below 65%	Overstaffed for current revenue. A decision is needed

Adjust for role. A senior lead who does estimates, reviews, and client management might have a target of 60 percent. A junior developer on a long project might run at 90. Set a target per role and compare each person to their own target.

Two patterns matter more than the average.

Concentrated bench. A team at 75 percent could be ten people each with a few spare hours a week, or seven people fully booked and nearly three with nothing. The first is slack. The second is a staffing problem with names attached, and usually a skills mismatch: the work you are selling no longer needs what those people do.

Persistent bench. A person benched for three weeks between projects is ordinary. The same person benched for three months is a signal that you have been avoiding a decision.

What to do with bench time

Bench time with no plan turns into anxious busywork. People know when they are not billing, and they draw conclusions about their job security. A prepared list protects the money and the person.

Keep a standing backlog of internal work, ranked, with an owner for each item and a definition of done. When someone comes free, they pick up the top item that fits their skills. In rough order of return:

Sales support. A developer who can spend two days on a technical approach for a live proposal raises your win rate and improves your estimate. This is the fastest route from bench hours back to billable ones. It also catches the under-scoped estimates that later become scope creep.

Short, paid work at flexible terms. A small discovery engagement, an audit, a fixed-price improvement for an existing client. A benched person billing at 70 percent of your usual rate is still far ahead of a benched person billing nothing. Keep two or three such offers ready to send.

Reusable assets. The deployment setup you rebuild on every project. The authentication module. The project starter with your standards built in. Each one shortens future projects, which improves the margin on fixed-price work.

Training tied to what you will sell. If the pipeline says clients are asking for AI features or a particular platform, bench time is when the team learns it. Tie the training to a concrete output, such as a working internal demo, so it finishes.

Documentation and handover notes. The work nobody does while billable, and the first thing you need when someone leaves.

Lending people out. Placing a benched developer with another agency or a client for a few months through staff augmentation keeps them billable. Price it properly, since the margin is thin.

One caution. Internal projects are a use for the bench and a poor reason to keep one. If a person’s time has been filled with internal work for a full quarter, be honest about whether the agency can afford that role at current revenue.

Forecasting the bench

The bench becomes a crisis when it arrives without warning. A capacity forecast gives you eight to twelve weeks of notice, which is enough time to sell, to reshuffle, or to make a harder decision calmly.

You need three numbers per month: capacity, signed work, and weighted pipeline. Weight each open opportunity by your honest estimate of its chance of closing. A 200-hour project at 50 percent counts as 100 hours.

For the same ten-person team, with 1,500 available hours a month:

	Month 1	Month 2	Month 3
Capacity (hours)	1,500	1,500	1,500
Signed work	1,250	950	600
Weighted pipeline	100	200	300
Forecast billable hours	1,350	1,150	900
Forecast utilization	90%	77%	60%
Forecast bench hours	150	350	600
Bench in full-time people (hours / 150)	1.0	2.3	4.0

Month 1 is too hot. Month 3 has four people’s worth of unsold time. Seen ten weeks ahead, that is a sales target: you need about 300 more hours of signed work in month 3 to reach 80 percent. Seen in the first week of month 3, it is $40,000 of salary cost with no time to fix it.

Run this forecast every week. It takes twenty minutes once the structure exists. Three habits keep it honest.

Count signed work only when it is signed. A verbal yes goes in the pipeline at a percentage.
Check your weights against history. If the opportunities you rated at 50 percent close a quarter of the time, halve your weights.
Forecast by skill as well as in total. Six hundred spare hours of front-end time does not help a pipeline full of data work.

A spreadsheet is enough for a team of ten. Past twenty people or a dozen concurrent projects, the spreadsheet starts to lag reality.

Acting on the forecast

A forecast is only useful if specific readings trigger specific actions. Agree on them in advance, while nothing is wrong.

Forecast reading	Action
Above 85% for the next two months	Line up contractors or augmented staff. Consider a hire if signed work covers twelve months
75% to 85%	Normal. Keep selling
Below 70% in month 3	Push sales now. Offer existing clients the small projects you have been deferring
Below 65% for two consecutive forecast months	Stop hiring and contractor renewals. Bring outsourced work back in-house. Review costs
Below 60% for a quarter with a thin pipeline	Run the runway arithmetic. This is the point where staffing changes come into view

That last row is where bench management meets its hardest outcome. If the numbers say the team is larger than the revenue can carry, the article on how to lay off employees covers how to check that conclusion and how to act on it properly. Owners who forecast get to that decision rarely, and when they do get there, they have had weeks to try everything else first.

The bench also argues for how you staff in the first place. Capacity you rent, through contractors, augmentation, or a partner who takes whole projects, disappears when the work does. Salaried capacity does not. The more uncertain the revenue, the more of it should sit on flexible capacity. The guide on when to partner instead of hire covers that tradeoff for specialist work.