There is a size at which an agency stops being limited by demand and starts being limited by its

founder. It is recognisable rather than measurable. Client work is good, referrals arrive, and yet every

account has a point where it waits for one person: the pack that needs their commentary, the pitch that

needs their voice, the difficult email nobody else will draft. The founder is not doing all the work by this

stage. They are doing the one piece in each workflow that everything else queues behind, which is a more constraining problem.

The usual response is to delegate by irritation. Owners hand over what they personally dislike, which

tends to mean administration, expenses, inbox tidying and diary management. Some of that is worth

doing, but as a first move it addresses the wrong thing. Disliking a task and being the bottleneck for it

are unrelated properties. Handing over the annoying half hour on a Friday leaves every queue where it

was, and produces the widely held view that support is nice to have rather than structural.

A more useful filter has three parts. Does the work repeat on a predictable rhythm, so somebody can get materially better at it. Does it have a knowable right answer, so quality is checkable without a debate

about taste. And does anything else queue behind it, so moving it actually shortens a path rather than

just tidying a desk. Work that passes all three is where delegation pays first, regardless of how the

founder feels about doing it.

Run the delivery workflow through that filter and the first thing to move is almost always assembly:

report packs built from dashboards, campaign builds executed from a plan, decks populated from a

skeleton, files renamed, resized and filed where the next person expects them. Assembly repeats, has a

right answer, and sits directly upstream of client-facing deadlines. It is also the work founders least

resent, which is exactly why it survives on their desk for years — it never irritates enough to be noticed

as the constraint it is.

The second move is the client communication layer, and it is the one owners flinch at. Not the relationship

— the layer. Status updates, meeting recaps sent the same day, approvals chased politely for the third

time, the note that warns a client a deliverable will land Thursday rather than Wednesday. None of

this is the difficult email; it is the connective tissue around it. Moved across, under the account lead's

name and to a standard the lead approves, it does something subtle: clients begin experiencing the

agency as responsive rather than experiencing the founder as responsive, which is the difference

between selling a firm and selling a person.

Judgement goes last, and some of it never goes. Pricing, the difficult conversation, the creative call

on which concept survives — these stay until there is someone whose judgement the founder has watched

at close range for months, and that is a hiring outcome, not a handover technique. The sequencing

error most owners make is trying to delegate judgement first, because it is the work that exhausts them

most, and concluding from the inevitable failure that nothing can be delegated at all.

Sequenced properly — assembly, then the communication layer, then judgement selectively — each

stage funds the next. Assembly frees the hours to document the communication layer; the communication

layer builds the trust that lets judgement move later; and somewhere in the middle, usually without

anyone announcing it, the agency stops queueing behind its owner.