Capacity and the owner bottleneck
Everyone said they had room. Two weeks later someone was at 140%.
An agency freelancer described the pattern better than we could:
“Everyone would say they had room, then two weeks in someone’s at 140% and the whole delivery schedule shifts. Sales would close deals and ops would scramble. Wasn’t hiring or churn. Just zero visibility on real capacity.”
This is the fast version of the margin problem. Margin blindness is chronic and costs you slowly. Capacity blindness is acute: it is why the quarter goes sideways in week three.
What changes if you fix it: on the sales call you know whether the work fits, because committed hours per person are a number rather than an opinion. Overload surfaces in week one instead of week three. And the approvals queued behind you shrink, because the ones that never needed you stop arriving.
The rest of this page is how that works, and what it costs.
Problem one: capacity is an opinion, not a number
The question gets asked as “do you have room?”, and it gets answered honestly by someone who is estimating their own future under social pressure, in a room where the answer everyone wants is yes.
That answer is then treated as a commitment. Sales closes against it. The schedule is built on it. And it was never a number — it was a guess made by a person who did not have the data to make it accurately and had no incentive to be the one who said no.
The failure shows up two weeks later, and by then three things have happened at once: someone is overloaded, another client’s work has slipped to accommodate it, and the account that slipped is now quietly forming an opinion about your reliability. That last one is expensive. Clients rank delivery dissatisfaction as the leading reason they leave — 48%, up fourteen points year on year — while agencies rank it seventh.
What we would do
Express every live retainer and project as committed hours per person per week, set against real availability, so a prospective deal is checked against a number before it is signed.
Real availability meaning: contracted hours minus holiday, minus the internal and admin load that genuinely exists, minus a deliberate buffer. Not a theoretical forty.
What changes: on the sales call, you know whether the work fits, and if it does not you know what would have to move or what the delivery date honestly is. Overload is visible in week one rather than week three, when it is still cheap to fix. And “we can start in three weeks” becomes something you can say with confidence — which sells better than an optimistic yes that turns into a slipped date.
Problem two: hiring moves work, it does not move approvals
The second problem is the one owners describe as being the bottleneck in their own business and then cannot solve by hiring.
You bring people in. The delivery load spreads. But the approvals do not — the sign-offs, the pricing calls, the exceptions, the “just check this with me before it goes”. Every new person and every new client adds decisions to the queue, and the queue has one server. Decision volume grows. The number of hours in your week does not.
The result is that everything waits on the same person, that person is in delivery all day, and the approvals happen in the evening. Growth makes it worse, not better.
What we would do
Here the honest answer is that there are two different situations and they need different work. Which one you are in is usually clear within an hour of looking at what actually reaches you.
If the decisions genuinely need you — pricing, scope exceptions, anything with real commercial consequence — then the target is not delegation, it is reducing the volume and cost of each decision. Batching them so they happen twice a day rather than continuously. Thresholds, so anything under a defined value proceeds without you. Default positions, so the common cases have a documented answer and only the exceptions surface. The decisions stay yours; the interruptions do not.
If they do not need you — and typically a good proportion do not, they arrive at you out of habit rather than necessity — then the work is routing: making it unambiguous who owns each class of decision, giving them what they need to make it, and defining the exception that comes back to you. This is slower to embed because it is a change in how people behave, not just in what a system does.
Most agencies have some of both. The first piece of work is finding out the split, because they are different projects with different costs.
What changes: the measurable version is how long work sits waiting on you, and how many approvals reach you in a week. Both should fall, visibly, and you should be able to see that they have.
A note on what this is and is not
This is operational plumbing. It is instrumentation, integration between the systems you already run, and putting a number in front of the person making the decision at the moment they make it. Some of it involves models and some of it emphatically does not, and we will not dress up a scheduling problem as something more sophisticated than it is.
We build it modularly and we build it to survive the tools changing, because they will. That is the whole methodology →
Tell us the job you would most like gone
One sentence by email. We will tell you straight whether software can kill it — and if it cannot, we will say so.
Or, if you would rather talk it through: book a discovery call. If you recognise the 140% week, that is the conversation.