The Agency Bench: Why Your Utilization Target Is Lying to You

Every agency owner I know has, at some point, written "target utilization: 85%" into a spreadsheet and felt like a serious operator. Then six months later a lead engineer quits, three projects slip at once, and nobody can explain why the P&L looks great but the team looks dead. The target wasn't wrong. The model behind it was.

This is a breakdown of how we think about utilization at 72Technologies — what the number actually measures, why the common targets are quietly destructive, and how to build a staffing model that survives real client behaviour.

What "utilization" actually means (and the three definitions people conflate)

When someone says "we run at 80% utilization," ask them which 80%. There are at least three numbers hiding in that sentence, and they drift apart fast.

• Billable utilization: hours billed to a client ÷ total paid hours. This is the one finance cares about.
• Booked utilization: hours assigned to a project in the staffing plan ÷ total paid hours. This is what your resourcing tool shows.
• Productive utilization: hours spent on client-facing work, billable or not (bug fixes under warranty, pre-sales scoping, pitch prototypes) ÷ total paid hours.

Agencies get in trouble when they set a target against one definition and measure against another. The classic failure mode: leadership sets an 85% billable target, resourcing books engineers to 95% to "leave buffer," and productive utilization quietly runs at 110% because warranty work, estimation, and interview loops don't appear anywhere.

A quick test

Ask your delivery lead: "If a senior engineer spends four hours on Thursday reviewing a proposal for a new client, where does that time land?" If the answer is "nowhere, they'll just log 40 billable hours anyway," you don't have a utilization problem — you have a measurement problem. The number on the dashboard is fiction.

The math nobody runs

Let's take the 85% target apart with a working year. Numbers vary by country; adjust to taste.

Total weeks in year:              52
- Public holidays (approx):       -2
- Paid leave:                     -4
- Sick / personal:                -1
- Training / internal:            -1
= Working weeks:                   44

Working hours (44 * 40):          1,760
85% billable target:              1,496 billable hours/year

So far, fine. Now layer in the things that actually happen in an agency:

• Pre-sales and scoping for prospects that don't convert
• Interviewing candidates (seniors do a lot of this)
• Code review across projects they're not booked on
• Warranty / post-launch fixes on finished projects
• The two-week gap between Project A ending Friday and Project B starting... whenever the client finally signs

In our experience, those line items eat somewhere between 10 and 20% of a senior engineer's year. If you've committed them to 85% billable, you've committed them to 95 – 105% total. That's the number that correlates with resignations, not the one on the dashboard.

Why high utilization destroys margin (yes, destroys)

The counter-intuitive bit: pushing utilization up past a certain threshold makes you less profitable, not more. A few reasons:

You lose the ability to say yes quickly. When every engineer is booked solid through Q3, the warm lead who wants to start next month gets told "we can slot you in October." They go elsewhere. The deal you lost was higher-margin than the one you're protecting.

You under-invest in sales motion. Pre-sales work done by engineers — technical discovery calls, architecture sketches, estimate reviews — is the single biggest predictor of win rate and project health that we've seen. If seniors are 85% billable, that work either doesn't happen or happens badly at 10 PM.

You pay for churn. Replacing a senior engineer in 2026 costs somewhere between three and nine months of their fully-loaded salary once you count recruiter fees, interview load, ramp time, and the delivery hit on their last project. One burnout-driven resignation can wipe out a year of "gains" from running the team 10 points hotter.

You can't absorb client chaos. Clients reschedule. Scope shifts. A key stakeholder goes on leave. If you've booked 95% of capacity, every one of those events becomes a crisis instead of a shrug.

A staffing model that actually holds up

Here's roughly how we structure capacity. Treat the percentages as starting points, not gospel.

Tier the roles, don't average them

Averaging utilization across the whole company is where strategy goes to die. Different roles should have different targets.

• Mid-level engineers: 75 – 80% billable. They're the delivery engine. Protect their focus time.
• Senior engineers / tech leads: 60 – 70% billable. The remaining 30 – 40% is pre-sales, mentoring, code review across projects, hiring, and architectural work on in-flight engagements.
• Principal / staff: 40 – 55% billable. If your principals are 85% booked to one project, you've turned your most expensive people into contractors. You're not getting the leverage you're paying for.
• Designers: 70 – 80%, with explicit non-billable blocks for design-system and portfolio work.
• PMs / delivery leads: measured in projects-owned, not hours. Utilization is the wrong lens.

Build a named bench, not an accidental one

An accidental bench is what happens when a project ends and nobody noticed. A named bench is a planned 10 – 15% of capacity that leadership knows is uncommitted next month, with a list of what it will do if no client work materialises:

1. Internal product work (if you're transitioning from agency to product, this is where it lives — see our services page for how we structure that split).
2. Deep R&D on a stack the sales team keeps hearing about.
3. Open-source contributions tied to your positioning.
4. Paid-down technical debt on a long-term retainer client.

If none of those exist, the bench will fill itself with Slack and dread. Have the list ready before you need it.

Price for the real target, not the aspirational one

This is the part most agencies skip. If your senior engineers realistically bill 65% of the year, your blended rate has to be built on 65%, not 85%. Otherwise every project is quietly subsidised by overwork.

Fully-loaded cost per senior/year:  $180,000  (salary + benefits + overhead)
Realistic billable hours:            1,760 * 0.65 = 1,144
Break-even hourly rate:              $157
Target margin (e.g. 2.2x):           $345/hour billed

If your current rate card was built assuming 1,500 billable hours and you're actually delivering 1,100, you're running a 25% hole and calling it "a tough quarter" four quarters in a row.

The conversation to have with your delivery lead this week

Two questions, in this order:

1. "What's the gap between booked and billable last quarter, by person?" If it's more than 10 points for anyone, you're either over-booking (and burning them) or under-tracking non-billable work (and lying to yourself about margin).
2. "Who did pre-sales work last quarter, and was it logged anywhere?" If the answer is "the usual two seniors, and no," you've found both your burnout risk and your margin leak in one sentence.

Neither question is about raising the target. They're about making the target honest.

Where we'd start

If you're running an agency and this landed uncomfortably close, don't rebuild the whole model this quarter. Do three things:

• Add a non-billable category called Pre-sales & scoping to your time tracker. Just that. Watch where the hours go for 60 days.
• Set role-specific targets instead of a company-wide one, even if the first version is rough.
• Recalculate your rate card against actual billable hours from last year, not planned ones. If the gap is uncomfortable, you've found your pricing conversation for 2026.

The goal isn't lower utilization for its own sake. It's a number you can trust — one that lines up with what your team is actually doing, so the decisions you make on top of it stop being guesses.