Operations
3 min readAgency capacity planning: how to know what one more client actually costs
What utilisation rate should an agency target?
Most agencies should target 65% to 75% billable utilisation for delivery staff, not 85% or higher. A 40-hour week does not contain 40 billable hours once internal meetings, admin, training, and rework are counted, so planning above 75% guarantees the overflow lands as unpaid overtime or dropped quality. Account managers and leads typically run lower, around 50% to 60%, because coordination is not billable but is not optional.
Every agency owner has signed a client they could not deliver. It usually happens in a good month, when the pipeline finally works and saying no feels insane.
A capacity model is what lets you say yes with a straight face.
The number nobody has
Ask an agency owner what one more $8,000 account costs to deliver, and you will get a shrug or an estimate. Ask what happens to delivery quality at eleven accounts versus nine, and you get a feeling.
Those are the two questions a capacity model answers, and the answers are arithmetic rather than instinct.
Build it in four steps
1. Hours per account, by role. For each service you sell, list the roles that touch it and the monthly hours each contributes. Use tracked hours from a real account, not an estimate. Estimates run 30 to 50% light, consistently.
2. Realistic utilisation per role. This is where most models break. A delivery specialist does not deliver 40 billable hours a week. Between internal meetings, admin, onboarding, training, and rework, 26 to 30 is realistic. That is 65 to 75% utilisation, and it is a healthy number rather than a slack one.
Leads and account managers run lower, 50 to 60%, because coordination consumes real time and is not billable. Planning them at 80% is how you lose your best people.
3. Capacity per person, per month. Multiply available hours by utilisation. A specialist at 30 billable hours a week has roughly 130 a month.
4. Divide. Total capacity by role, divided by hours per account by role, gives you maximum accounts. The binding role, the one that runs out first, is your delivery constraint. It is rarely the one people assume.
What the model tells you that a gut feel does not
Whether to hire before you sell. If you are at 92% of capacity on the binding role, the next account needs a hire, and hires take 60 to 90 days to become productive. That is a decision to make in advance, not in week two of a signed contract.
Which service line is quietly killing you. Hours per dollar varies wildly across services. Once you can see it, the service you thought was your flagship is sometimes your worst.
What a headcount actually buys. "Can we afford another specialist" becomes "a specialist adds capacity for 2.3 accounts at $8,000, so $18,400 of monthly revenue against $7,100 of fully-burdened cost." That is a decision. The other version is a worry.
Utilisation is a planning number, not a target
One warning. The moment utilisation becomes a performance metric people are judged on, it stops being true. Hours get recorded to hit the number and your model quietly fills with fiction.
Track it to plan capacity. Judge people on delivery quality and client outcomes. Keep those two things apart.
Where it connects
Capacity planning and pricing are the same exercise viewed from two directions. The capacity model tells you what a month of delivery costs; pricing turns that into a floor. Run them together or neither one is real.
By Nicholas Kirchner · Updated August 5, 2026
Read next