The question always arrives the same way. A founder is looking at a resourcing spreadsheet, a new client is about to sign, and someone asks: can Sarah take one more?
Nobody knows. So they guess. Sarah is competent and agreeable, so she says yes. Four weeks later a deadline slips on an unrelated account, and everyone is surprised.
Before starting Scopeyard, I spent years running a product development studio and delivering AI automation projects across healthcare, recruitment and operations. I have made this exact mistake, more than once. And the reason it keeps happening is that "how many projects can one project manager handle" is the wrong question. Projects are not a unit of load. A four-week landing page and an eighteen-month platform build both count as one project, and one of them will eat a PM alive.
The right question is how much coordination surface a person can hold before things start falling through it.
PM Capacity = Available coordination hours ÷ Coordination load per project
Everything below is how to work out both sides of that division.
1. Start with what the benchmarks actually say
There is data here, and it is worth knowing before you invent your own number.
Across general project management survey data, roughly 59% of project managers run two to five projects at once, 11% run six to ten, 15% run more than ten, and 15% run a single project. So the modal answer is three-ish, with a long tail of people managing portfolios of small work.
Staffing ratios tell a similar story from another angle. The commonly repeated "one PM per eight delivery staff" benchmark tends to break in practice; firms that have scaled report 1:7 as the upper edge of still-functional, with 1:4 to 1:5 far more sustainable, and tighter still — around 1:5 — for distributed teams and contractor-heavy setups where coordination overhead is higher.
And utilisation benchmarks quietly explain why. Agency-wide utilisation sits around 65% in UK benchmark data, with delivery staff — designers, developers, copywriters — expected at 80–85%, while PM and account roles sit at 55–70% billable. That gap is not slack. It is the coordination work: status, chasing, internal syncs, scope conversations, the bits that never appear on a timesheet line a client will pay for.
| Signal | Typical benchmark | What it means for you |
|---|---|---|
| Projects per PM | 2–5 (59% of PMs) | Three is normal, not impressive |
| PM to delivery staff | 1:4 to 1:7 | Beyond 1:7 you are relying on heroics |
| PM billable utilisation | 55–70% | 30–45% of the role is coordination, by design |
| Delivery staff utilisation | 80–85% | Do not benchmark PMs against this |
| Sustained utilisation above 85% | Warning sign | Usually means absorbed scope or chronic underestimating |
Use these as sanity checks, not answers. A PM running five retainers with fixed monthly rhythms is not carrying the same load as a PM running five simultaneous discovery phases.
2. Work out how many coordination hours you actually have
Take one PM. Thirty-eight hours a week. Now subtract honestly.
Internal meetings, one-to-ones, hiring, admin and the general tax of working inside an organisation take 6–8 hours. If your PM is also doing pre-sales scoping — and in most agencies under thirty people, they are — that is another 3–5 hours in a busy month. What is left, call it 25 hours, is the real coordination budget.
That number is the numerator in the equation, and it is smaller than anyone's spreadsheet assumes. Most resourcing plans quietly allocate 38 hours of a 25-hour person.
Then apply the discount nobody applies. Research summarised by the American Psychological Association puts the cost of chronic multitasking and context switching at up to 40% of productive time. Asana's Anatomy of Work research puts it differently but lands in the same place: knowledge workers spend around 60% of their time on "work about work" — chasing status, duplicating effort, sitting in meetings about the meeting — with 352 hours a year going to check-ins and status updates alone.
So a PM on four projects does not have 25 hours split four ways. She has something closer to 15 usable hours, because the switching itself is billed to her.
3. Price the coordination load per project
Now the denominator. Not every project costs the same to hold. Score each one honestly — I use four factors, all of which map to how often someone has to make a decision or chase one.
- Stakeholder count. One decision maker or six? Every extra approver is a queue.
- Novelty. Have you shipped this exact shape of work before, or are you inventing the plan as you go?
- Client maturity. Does the client respond within two days and know what they want, or do they go silent then change their mind?
- Delivery volatility. Fixed monthly retainer work, or an integration-heavy build where any dependency can blow up the week?
Here is roughly what those archetypes cost in coordination hours per week, based on how we planned resourcing at the studio and what I have seen since.
| Project archetype | Coordination hours / week | Realistic per PM |
|---|---|---|
| Mature retainer, one stakeholder, repeatable scope | 2–3 | 6–8 |
| Standard website or campaign build, clear scope | 4–6 | 4–5 |
| Product build, multiple stakeholders, active discovery | 8–12 | 2–3 |
| AI automation build with data access and eval cycles | 10–15 | 1–2 |
| Anything in escalation or recovery | 15+ | 1, and nothing else |
Two things fall out of this table immediately. First, the honest answer to "how many projects can one PM handle" is between one and eight depending entirely on what kind of projects they are. Second, mixing archetypes is where teams get hurt — a PM with one recovery project and three "easy" retainers has zero capacity, not three-quarters of a person spare.
AI work deserves a specific note. Automation projects carry more coordination per pound of build than almost anything else, because they run on data access nobody has, edge cases nobody documented, and evaluation cycles that need a human to sit and judge outputs. I have written about why in how to scope AI automation projects, but for resourcing purposes: assume an AI build costs a PM roughly double what a website of the same contract value does.
4. Fix the multipliers before you change the number
If a PM is at capacity, the reflex is to hire another PM. Sometimes that is right. More often, the coordination cost per project is inflated by things you control.
Fixed cadence beats ad hoc. A weekly delivery meeting at the same time every week, with the same agenda, costs less than the same information gathered through fourteen improvised Slack threads. Rhythm is cheaper than responsiveness.
One source of truth. The moment status lives in a project tool and email and WhatsApp, your PM becomes the sync layer between them, and the sync layer is a full-time job that produces nothing. This is the single biggest inflator I see.
Templates for anything repeated. If you deliver the same shape of project more than three times a year, 70% of the plan should exist before kickoff. A library of SOPs is not bureaucracy; it is the difference between a PM planning and a PM re-planning.
Structured approvals. Chasing sign-off is the highest-frequency, lowest-value activity in the PM role. Every approval that needs a human to remember, remind and record it is a recurring tax — and PMI's benchmark that organisations waste around 11.4% of project investment to poor performance is largely made of exactly this kind of leakage.
Improve those four and the same person genuinely holds an extra project or two. Ignore them and hiring another PM just gives you two people drowning instead of one.
5. Watch for over-capacity before the client tells you
Over-capacity does not announce itself. It shows up as small, specific behaviours, usually three to four weeks before a deadline slips:
- Status updates go out late, or stop being written at all
- The PM starts answering questions from memory instead of from the board
- Decisions get deferred to "let me check and come back to you" and never come back
- Internal reviews get skipped so client-facing work stays on track
- The PM is fast on whatever is loudest and slow on everything else
That last one is the clearest signal. A PM at capacity triages by volume, not by importance. When you notice it, you are already past the line.
The blunt test I use: ask the PM, without warning, what the next milestone and named approver is on each of their projects. If they can answer for all of them in under a minute, they have room. If they need to open three tools, they do not.
6. When you are over, choose deliberately
There are exactly four options, and "try harder" is not among them.
Reduce the number of projects. Reduce the coordination load per project by fixing cadence, tooling and approvals. Add a coordinator underneath the PM — a junior owning scheduling, notes and chasing, which is usually cheaper and faster than a second PM. Or slow intake, which nobody wants to hear and is sometimes the only honest answer.
At Scopeyard we built the delivery layer around the third and fourth of the multipliers above — milestone-based views clients understand, feedback and approvals recorded against each deliverable, unlimited client access so you are never rationing seats — because the majority of the coordination hours we measured were a human being manually holding state that a system should hold. If you run automation work, the AI agencies page shows how those pieces fit; if you want the practical habits without the tool, managing multiple client projects without chaos covers them.
Final thoughts
Most agencies answer this question with a number they inherited: three, or five, or "as many as she can". Then they staff to the number and wonder why delivery feels fragile.
Work it out instead. Real coordination hours on top, honest load per project underneath, and the four multipliers applied before anyone gets hired.
A project manager's capacity is not a personality trait. It is a measurement of how much of your delivery system they are being asked to carry in their head.
Sources: Indeed — How Many Projects Should Project Managers Oversee and Manage?, Synergist — Agency utilisation guide, Haus Advisors — Agency Utilization Rate: Benchmarks by Role, Asana — The Anatomy of Work, PMI — Pulse of the Profession.