Every agency needs four functions covered regardless of size: winning work, delivering it, managing the client relationship, and running the business. Below about eight people those are hats rather than roles, and the useful question is not "what is our org chart" but "who owns each function today, and who covers it when they are away". The split that matters most as you grow is account management from project management - one owns the relationship and the commercial outcome, the other owns the timeline and the resource - and combining them in one person is the most common structural cause of both scope creep and burnout.
Agency org charts are usually drawn too early or too late.
Too early, and a six-person business acquires titles and reporting lines that describe an organisation it does not have, which makes everyone's job smaller than it needs to be. Too late, and a twenty-person agency is still operating as though everyone can hold everything in their head, with the founder as the single point through which all decisions pass.
The useful framing at any size is not the chart. It is: which functions must be covered, who covers each one today, and what happens when that person is away?
This guide covers the four functions, the roles that emerge as agencies grow, the account-versus-project-manager split, the three common structures, and the signals that tell you the current shape has stopped working.
The four functions
Every agency, from one person to two hundred, needs these covered.
Winning work. Positioning, marketing, sales, proposals. Someone owns whether there is a pipeline.
Delivering work. The craft itself, plus the machinery around it - planning, sequencing, quality. Someone owns whether it ships.
Managing the relationship. Client communication, expectations, scope, renewal. Someone owns whether the client is happy and whether they come back.
Running the business. Finance, hiring, tooling, legal, strategy. Someone owns whether the business is viable.
In a one-person agency all four are the same person. At three people they are informally split. At ten they need naming, because the failure mode at ten is that everyone assumes someone else is covering something.
The diagnostic worth running today: write the four functions down and name who owns each. If any function has no name, or has three names, that is where your problems are coming from.
The roles, and when they appear
Workamajig's overview of agency titles and Saka & Company's breakdown of the six core agency roles both map the standard shapes. The sequence below is the order they typically become necessary.
Delivery specialists first. Designers, developers, writers, strategists. The first hire is almost always here, because production capacity is the binding constraint in most small agencies. See hiring your first employee.
Project management second, usually around six to eight people. The signal is that coordination has become somebody's unacknowledged second job and is being done badly because it is nobody's first.
Account management third, typically eight to twelve. The signal is that client relationships are being handled reactively - nobody is thinking about renewal, growth or satisfaction until something goes wrong.
Operations fourth. Finance, resourcing, tooling, process. Often part-time or fractional long before it is full-time, and frequently the role a founder should hire to replace themselves in.
Discipline leads later, as teams grow past the point where one person can maintain quality and develop people across everything.
Account manager versus project manager
The split that matters most, and the one most often collapsed into a single role.
The project manager owns the work. Timeline, resource, dependencies, delivery. Their question is "will this ship on time, to scope, within budget?"
The account manager owns the relationship. Client satisfaction, commercial outcome, scope negotiation, growth and renewal. Their question is "is this client getting what they need, and will they still be here next year?"
Workamajig's comparison of the two roles puts it simply: one manages internally, the other manages the client, and together they cover what neither can alone.
Why combining them causes problems. The two roles have a productive tension. The account manager wants the client happy; the project manager wants the scope held. When one person holds both, the tension resolves internally and almost always in the same direction - toward the client, because that pressure is immediate and visible while the margin pressure is abstract and later.
That is a structural explanation for two problems agencies usually treat as cultural: scope absorbed without a change order, and delivery teams absorbing the consequences. It is not that the person is weak. It is that you asked one person to argue both sides.
Below eight people you cannot afford the split, and the mitigation is to make the tension explicit - the person wearing both hats should say out loud which one they are wearing when a scope decision arises.
Three structures
Functional
Teams organised by discipline - all designers together, all developers together - with people assigned to projects as needed.
Good for: craft development, consistent standards, flexible resourcing. Bad at: client continuity. Clients meet a rotating cast, and context is re-explained constantly. Fits: agencies with many short projects and a strong craft identity.
Pods
Small cross-functional teams that own a set of clients end to end - a strategist, a designer, a developer, a project manager.
Good for: client continuity, accountability, speed. The pod knows the client, so nothing is re-explained. Bad at: utilisation efficiency. A pod with a quiet month has idle capacity that is awkward to lend out, and specialist skills get duplicated across pods. Fits: retainer-heavy agencies and longer engagements. Saka's overview of team structures covers the trade-offs across the common models.
Hybrid
Core pods for continuity, with specialists shared across them.
Good for: most agencies between ten and forty people. Bad at: clarity. Shared specialists have two masters, and without explicit prioritisation they become the bottleneck everyone waits on. Fits: growing agencies, provided someone owns the shared resource allocation - which is where capacity planning stops being optional.
There is no correct answer. The trade-off is genuinely between continuity and utilisation, and which one you should optimise for depends on whether your revenue is project-based or retainer-based.
The signals that the current shape has stopped working
Six things worth watching for.
Everything routes through the founder. The clearest signal, and it is a bottleneck rather than a compliment. If decisions wait for one person, the business cannot grow past their calendar.
Nobody owns the client between projects. Retainers drift, renewals surprise you, and growth opportunities are invisible. This is the account management gap.
Deadlines slip for coordination reasons rather than capability reasons. The work is good and the sequencing is not. This is the project management gap.
The same question is answered differently by different people. Process exists in individual heads rather than anywhere shared.
Specialists are permanently the bottleneck. One person that every project needs. Either a hiring problem or a prioritisation problem, and it is worth diagnosing which before spending money.
Nobody is looking at the numbers. Utilisation, margin, cash, pipeline. This is the operations gap and it usually appears before anyone names it.
Structure at each size
Rough shapes, not prescriptions.
1-3 people. No structure. The founder holds all four functions; specialists deliver. The risk is that business-running gets no time at all, because it is the only function with no external deadline.
4-8. Functions still held as hats, but named. One person should own delivery coordination even if it is 30% of their week. The founder should be actively trying to hand over one function entirely - usually delivery coordination, sometimes finance.
9-15. Roles rather than hats. Project management is a real job. Account management is emerging. Operations is at least fractional. The founder should be out of day-to-day delivery, which is the transition most founders find hardest and delay longest.
16-30. Pods or hybrid. Discipline leads appear. Account management and project management are genuinely separate. The founder is running the business rather than the work.
The transitions are uncomfortable in a predictable way: each one requires the founder to stop doing something they are good at, in favour of something they are worse at and enjoy less.
The founder transitions
Agency structure changes are really a sequence of things the founder stops doing, and each one has a recognisable failure mode.
Stopping delivery. The first and hardest. A founder who is the best designer in the building has to accept work being done differently and initially worse, or the business cannot grow past their personal capacity. The failure mode is keeping the interesting projects and delegating the dull ones, which produces a team that never develops judgement and a founder who is still the bottleneck.
Stopping day-to-day client contact. Clients bought the founder, or believe they did, and handing over feels like a downgrade. It is manageable with a proper transition - overlap for a few weeks, explain the change positively, and stay reachable for escalation. The failure mode is a half-handover where the client keeps calling the founder anyway, which undermines whoever supposedly owns the account.
Stopping being the approver on everything. The subtlest one. A founder who reviews every deliverable is a queue, and the queue lengthens as the business grows. The transition requires naming who approves what and then genuinely not overriding it, which is harder than it sounds the first time you disagree with a decision.
Stopping doing the finance and admin. Usually the last, often the one that should have been first. A fractional operations or finance person is frequently the highest-return hire an agency of ten makes, and it is almost always delayed because it does not feel like it adds capacity.
Each transition has the same shape: the founder gives up something they are good at, in exchange for the business being able to do it without them. That trade feels bad in the moment and is the only route to a business that is worth more than its founder's calendar.
Structuring for retainers versus projects
The revenue model should drive the structure more than it usually does.
Retainer-heavy agencies benefit from continuity, which argues for pods or named account teams. A retainer client is buying responsiveness and accumulated context; rotating people through them destroys exactly what they are paying for. The cost is utilisation efficiency, and it is usually worth paying.
Project-heavy agencies benefit from flexibility, which argues for a functional structure with people assigned as needed. Projects have defined ends, so continuity matters less and the ability to staff a spike matters more.
Mixed agencies - which is most - need to decide which side to optimise for rather than splitting the difference badly. A common workable answer is a small core team dedicated to retainers, with project work staffed from a shared pool. That gives the retainer clients continuity and keeps the project side flexible, at the cost of some duplication.
What does not work is running a functional structure while selling retainers on the promise of a dedicated team. The mismatch surfaces within a quarter, usually as a client complaint about "never speaking to the same person twice."
Governance without bureaucracy
Beyond the four functions, three decisions need an explicit owner as an agency passes ten people. They are the ones that otherwise get made by whoever happens to be in the room.
Who approves a discount. Without a named owner, discounting spreads, because it is always locally rational to win the deal in front of you. Our guide to pricing agency services covers why this quietly resets your rate across the client base.
Who decides whether to take on a client. Declining work is a positioning decision and it needs authority behind it, or you will accept everything and discover the consequences in the project margin spread.
Who owns process changes. After a retrospective produces three actions, someone has to make them real. Unowned process improvements do not happen, and after two cycles of that the team stops taking retrospectives seriously.
Three names, written down. That is the whole governance layer a twenty-person agency needs, and it prevents more problems than any org chart.
Titles, and how much they matter
Less than people fear internally and more than people expect externally.
Internally, titles at a small agency are mostly noise. What matters is the delegation boundary - what someone decides alone, checks, or escalates - and that is a conversation rather than a label.
Externally, titles carry real weight. Clients read them as signals of seniority and authority, and a client who believes they are speaking to someone junior will escalate around them. That is a genuine argument for client-facing titles that reflect actual authority rather than tenure.
Two practical rules. Do not inflate titles you cannot sustain - a four-person agency with three directors has nowhere to go and looks odd to anyone who counts. And do not give someone a client-facing title without the authority to match, because they will be asked to make decisions and having to check every one undermines them with the client permanently.
Common structural mistakes
Five patterns that show up repeatedly, each with a recognisable cost.
Hiring a senior person into an unstructured business. A senior hire brought in to "bring order" without authority, budget or a defined remit will spend six months negotiating for permission and then leave. Seniority without authority is expensive and demoralising for everyone.
Creating a role to solve a person problem. Promoting someone into a management position because they are good at the craft, or inventing a role to retain someone, produces a structure shaped around individuals rather than needs. It works until that person leaves, at which point nothing fits.
Splitting account and project management too early. Below roughly eight people you cannot afford two roles where one person's time would do, and doing it anyway means both are underutilised and the coordination overhead between them exceeds the benefit.
Adding management before adding capacity. A layer of coordination over a team that is already at capacity does not increase throughput; it increases the number of people discussing the throughput. The check is whether delivery output rises within two quarters.
Never revisiting. The most common of all. A structure that fitted at eight people is quietly wrong at eighteen, and because the change is gradual nobody names the moment it stopped working. An annual look at the four functions and who owns them is enough to catch it.
Structure and profitability
Worth connecting explicitly, because structure decisions are usually made on feel and they show up in the numbers.
Every non-billable role you add raises the utilisation the billable roles must achieve to sustain the same margin. A project manager who is 20% billable has to be paid for by everyone else, which is entirely fine when they raise delivery throughput by more than their cost - and is not fine when they were hired because coordination felt chaotic and nobody measured whether it improved.
The check is straightforward: after adding a non-billable role, does gross margin hold within two quarters? Our guide to agency financial metrics covers the numbers to watch. If margin falls and stays down, the role either is not paying for itself or the rest of the structure needs to grow into it.
Documenting it without bureaucracy
You do not need an org chart. You need three things written down.
Who owns each of the four functions. One line each.
Who deputises. For every role, who covers it during absence. The most common operational failure in a small agency is one person's holiday stalling three projects, and it is entirely preventable with one sentence per role.
What each role decides alone. The delegation boundary. What they decide, what they consult on, what always escalates. Ambiguity here produces either paralysis or unpleasant surprises, and one conversation prevents both.
That is a page. It does more practical work than any chart, because it answers the questions people actually have.
A worked example: eight to sixteen people
The transition most agencies find hardest, laid out concretely.
At eight, a typical shape is a founder, four or five delivery specialists, one person doing project coordination alongside delivery, and outsourced bookkeeping. The founder holds sales, account management and final approval on everything, and is still delivering perhaps 40% of their time.
The strain shows in three places: the founder is the approval queue, nobody owns clients between projects, and coordination is being done in the gaps by someone whose main job is something else.
The first change is usually to make project coordination a real role rather than a side task. That is a genuine hire or a genuine internal move with the delivery work reassigned - not a title added to someone's existing job, which changes nothing.
The second is the founder stopping delivery. This is the transition that unlocks the rest, and it is almost always delayed by a year. Until it happens, the founder cannot take on account management properly, cannot sell consistently, and remains the bottleneck on approvals.
The third, around twelve, is account management becoming distinct. Often this is the founder handing over their smaller accounts rather than a new hire, which is cheaper and works better because the relationships transfer with someone who knows them.
The fourth, around fourteen to sixteen, is fractional operations or finance. By this point the administrative load - resourcing, invoicing, contracts, tooling, hiring - is a real job being done badly by people whose time is worth more elsewhere.
By sixteen you have four functions with named owners, a founder who runs the business rather than the work, and enough structure that a person going on holiday does not stall three projects. That is the whole objective, and the sequence above is the order that usually works.
The honest summary
Structure is not about titles. It is about making sure the four functions are covered, that somebody owns each, and that the coverage survives someone being on holiday.
The two changes with the largest effect are separating account management from project management once you can afford it, and getting the founder out of day-to-day delivery - in that order, and both later than they should have happened in almost every agency that has done them.