When Your Agency Needs a COO or Operations Lead

The question usually arrives shaped as a size question. How big does an agency have to be before it needs a COO. It is the wrong question, and asking it that way is how agencies end up hiring this seat a year late or two years early.

I sell this seat, so treat what follows as interested testimony and check it against your own numbers. The record behind it is twelve years running delivery operations for marketing agencies and 92 completed engagements. This piece covers one decision only: whether your agency needs an operations lead at all, and if so, when. Which seat, and how to hire into it, are separate questions I will point you at. I want to stay on the prior question, because it is the one most often skipped.

Does your agency need a COO, or does it need more clients?

Start with the fork that decides everything downstream. Is your constraint demand or delivery?

If new business is the constraint, an operations lead will make your agency calmer and no larger. Operations work turns a chaotic full pipeline into a predictable one. It cannot fill a pipeline that is not full. Hiring for order during a quiet quarter buys an expensive review of a machine that is not under load.

If delivery is the constraint, it shows up in one of three ways. You are turning work away. You are delivering late. Or you are delivering on time by absorbing the difference personally. The third is the dangerous one, because it appears in no report anywhere. The agency looks healthy and the founder is the shock absorber.

What actually breaks when an agency outgrows one operator?

Not the work. The coordination around the work.

Every agency runs on decisions that have to be made continuously. Who takes the new project. What slips when two deadlines collide. Whether this request sits inside scope. Which client gets the senior person this month. In a small agency those calls are cheap, because one person holds all the context and can make them in seconds.

Size does not make any single decision harder. It multiplies how many there are, and it scatters the context needed to make them well across several heads. The founder still makes the calls, now on partial information, and increasingly makes them late, because the queue has become longer than the day.

That is the real failure mode. Not incompetence, not bad tooling. Decision volume outgrowing one head. Naming it matters, because the instinct is to fix it by hiring more delivery people, and every delivery hire adds coordination load to the same overloaded head. It is why an agency that grew from eight people to fourteen often feels worse at fourteen.

What are the signs your agency needs an operations lead?

Six symptoms. Any one of them alone is noise. Three or four at once is a pattern.

• Adding people stopped adding output. You hired, the load did not drop, and you cannot fully explain why. The capacity is real. It is being spent on coordination that did not exist before.
• The same mistake reaches different clients through different people. One person missing a step is a person. Three people missing the same step across three months is a system that never specified the step, and nobody owns the system.
• Two people give the same client two different answers. Status, dates, what was agreed in the last call. When the truth lives in individual heads rather than one place, your internal disorder becomes the client’s experience.
• Scope creep is discovered at invoicing. The extra work happened weeks earlier. If nobody caught it at the moment it was requested, there is no checkpoint in the process where scope gets defended, and margin leaves quietly.
• Quality tracks how busy you personally are. Clean work in a slow month, sloppy work in a heavy one. That means you are the quality control layer, and you are a layer that fails precisely when it is needed most.
• You have stopped selling because you are delivering. This is the symptom that compounds. Delivery pressure eats the founder’s business development time, the pipeline thins a quarter later, and the quiet quarter lands just after you finally caught up.

At what size does an agency need a COO?

Headcount is a proxy, and a poor one, but people want a number, so here is the honest version of it.

Under about ten people, most agencies do not have an operating layer large enough for anyone to own. The pain is usually project-level: dates, status, client updates, coordination between busy people. That is a project manager, and a more senior seat will not solve it faster.

Between roughly ten and fifty, the operating layer exists as real work but rarely as a full week of it. This is the range where the need is genuine and the shape of the answer is open, which is the whole reason fractional and consulting versions of this seat exist.

Past fifty, whether is no longer the question. If nobody owns operations at that size, you already know it.

Two things move those thresholds more than headcount does. The number of concurrent client projects, and how many different services you sell. An agency of twelve running one repeatable service for twenty retainer clients carries less operating surface than an agency of eight selling four services with custom scopes. Count workflows, not desks.

Does the seat pay for itself?

I am not going to give you a salary benchmark, and not out of coyness. A market rate tells you what other agencies pay. It does not tell you whether this seat returns more than it costs inside your agency, and that second question is answerable without a benchmark at all.

The return, where it exists, comes from four places.

• Founder hours, valued at their best alternative use. An hour you spend rescheduling a project is not worth an hour of a coordinator’s time. It is worth whatever you would otherwise have done, usually the pitch you did not prepare.
• Margin that stops leaking. Unbilled scope, rework and projects that quietly run long all come out of margin. Those are operating failures rather than pricing failures, and raising your rates does not fix them.
• Capacity you already own but cannot see. Plenty of agencies are not at capacity. They are at the limit of their visibility, and that is a far cheaper problem to solve.
• Retention. The client losses I have watched up close were rarely about the quality of the work. They were about an experience that felt unpredictable, and predictability is manufactured in operations.

If you cannot see a plausible return in at least two of those four, the seat is premature. That test holds at any price point, which is what makes it useful.

When should you not hire one yet?

Four situations where the answer is no, or at least not this.

• Demand is the constraint. Covered above, and still the most common false positive.
• The pain is project-level. If what hurts is dates, status and chasing, hire a project manager and give them real authority. An operations lead is a more expensive way to be disappointed by the same gap.
• One unresolved personnel problem is wearing a systems costume. If everybody in the company can name the person, no amount of process design substitutes for the conversation you have been avoiding.
• You have not settled what the agency sells. An operations lead can make a defined offer run reliably. They cannot decide what the offer is. Systematising a service menu that changes with every proposal produces a system nobody follows.

Can you promote someone internally instead?

Often yes, and this is the option owners dismiss too quickly. Someone already inside has the context an outsider spends a quarter acquiring. They know which client is fragile, which project always runs long, and which promises your team has learned not to make.

The case against is what sinks most internal promotions. You are usually promoting your best delivery person out of delivery, and the operating layer is a different job rather than a senior version of the same one. Running projects well does not mean being able to design the system projects run inside. Finding out costs you the system and the project manager at the same time.

The version I have seen work is a split. Buy the design work once from outside, then hand the ongoing seat to your internal person, who inherits a documented operation instead of inventing one under live client load.

If you have decided the need is real, the next two questions are which seat and how to fill it, and both have their own answers: the difference between a COO, an operations consultant and an integrator, and a hiring guide covering vetting, decision rights and the first 90 days in how to hire a fractional COO. What the ongoing seat covers week to week is set out on my fractional COO services page.

So when does an agency actually need a COO?

My verdict, with my interest in it already on the table. An agency needs an operations lead when decision volume has outgrown one head, not when headcount crosses a line. For most marketing agencies that lands somewhere between ten and thirty people, earlier if you sell several custom services, later if you sell one repeatable service. Below it, the honest answer is a project manager and a defined process. Past fifty without one, you are late.

The second half of the verdict is the part that costs people money. Act the quarter before it hurts, not the quarter it does. Operations work compounds slowly, so the payoff lands in the quarter after the one you buy. Hire at peak pain and relief arrives late, and then the seat gets judged on how the worst quarter went. If three or four of those six symptoms are true this week, this is a decision to make rather than a trend to watch.

Where to start if you are not sure

You do not have to make this call on instinct. The first step I recommend is the Agency Ops Audit: $1,500 fixed, two weeks, a written 90-day roadmap and a 60-minute readout call. It is recommended, not required: a retainer can also begin after a scoping call. It answers the whether question against your own delivery, not a benchmark. If the finding is that you need a project manager, or that demand is the real constraint, or that nothing should be hired for two more quarters, that is what the roadmap will say. I would rather tell you that than sell you a quarter of the wrong seat.