Agency pod structure: how to grow by cloning a unit instead of redesigning the org chart

Most agencies grow by adding people to a pool. A new client signs, someone notices the team looks stretched, and a hire goes out. The work is distributed by whoever has bandwidth that week, which is usually decided in a Monday meeting by the person with the loudest problem.

That model works to about fifteen clients. Past that it produces a specific set of symptoms: nobody can say what one more client costs, the best people are quietly carrying the worst accounts, and every new hire makes the org chart harder to explain rather than easier.

What tends to resolve all three at once is changing the unit you grow in, so that adding capacity means adding another complete unit rather than adding another person to a shared pool.

What a pod actually is

A pod is a fixed roster of delivery roles with a known client capacity, run as one repeatable unit.

The important word is fixed. A pod is a named set of seats, the same every time, sized so it holds a specific number of accounts at the quality you sell. "Roughly five people, depending" is a pool with better branding. Once you know what one pod produces and what one client consumes, capacity stops being a feeling and becomes a number you can count.

Growth then means staffing another pod and leaving the ones you already have alone.

Sizing the first one

Size the pod against what one account consumes. Your current headcount answers a different question, which is who you happen to have.

Start with what one account actually consumes. The quoted scope was a guess made before the work happened. Pull the last three months and count the real hours: delivery, revisions, the account call, the internal prep for the account call, and the work that gets redone because the brief was thin. Most agencies are surprised by a factor of about 1.4.

Then set the capacity per seat honestly. A forty-hour week does not contain forty billable hours. Sixty-five to seventy-five percent utilisation is the realistic band for delivery staff, and leads run lower because coordination is not billable but is not optional. Plan above that and the overflow lands as unpaid overtime or as quality you did not intend to sell.

Now the roster falls out of the arithmetic. Client hours divided by realistic seat capacity gives you the seats. If the answer is 4.3 people, round up. Five seats give you headroom. Four seats give you a deficit that gets paid in overtime.

Last, check the margin. Fully burdened cost of the pod against the revenue it can hold at full capacity. If that does not clear your target gross margin with the pod full, it will never clear it with the pod half full, and no amount of volume fixes a unit that is unprofitable by design.

Why this changes what growth feels like

The benefit that compounds over time is that the decisions get smaller, because each one is now bounded by a unit rather than by the whole business. A pool-based agency answers "can we take this client?" with a judgement call that depends on who is asked. A pod-based agency answers it by looking at which pod has a seat. When every pod is full, the answer is either "we hire the next pod" or "we raise prices until demand matches capacity", and both of those are decisions an owner can make from a number rather than from a feeling.

It also makes the ceiling visible in advance. You can see what the business looks like at three pods and at five pods, including what it costs to get there, before spending anything. Most agencies discover their next constraint by hitting it. This is how you see it coming.

Where it goes wrong

Pods that are not really fixed. The moment one pod borrows a designer from another to cover a deadline, you are back to a pool with extra vocabulary. If borrowing is happening weekly, the pod is undersized and the honest fix is to resize it, not to keep lending.

Sizing to your best people. If the pod only works when the strongest editor is in it, what you have built is a dependency on that editor. Size to the standard you can hire to, repeatedly.

Cloning before the first one works. A pod that is not hitting its margin at capacity does not get better when duplicated. It gets duplicated. Prove the unit, then clone it.

Forgetting the layer above. Two pods coordinate themselves. Five do not. Somewhere around the third pod you need someone owning across pods, and that seat has to be costed into the model rather than absorbed by whoever notices first.

Where it connects

Pod structure is the delivery-side answer to the same question pricing answers from the commercial side: what does one more client actually cost, and does the price cover it. Get the capacity model right and pricing becomes arithmetic. Get it wrong and every retainer is a guess that happens to have a number on it.

Read next

• Why agency growth eats margin, and how to price so it doesn't
• Agency capacity planning: how to know what one more client actually costs