Managing People, Not Just Headcount

Managing People, Not Just Headcount

Why capacity is not a number, and what your best people stop doing before they ever miss a deadline

The most expensive sentence I have ever said was one I said to myself.

They can handle one more project.

It was early in my time running operations at scale. We had several client launches happening at once, which is a normal state of affairs in this industry and never stops feeling abnormal. I was looking at allocation and one name stood out for the obvious reason. They were excellent. They delivered consistently. They almost never complained. They were, in the most useful sense of the word, reliable. So I gave them one more thing.

On paper it was the correct decision. If you had shown me the same spreadsheet a second time, with the same information, I would have made the same call. Their availability supported it. Their track record supported it. Nothing in the data said no.

What happened next did not show up for weeks, and when it did show up it did not look like a problem.

Their replies got slower. Not late, just slower. Then they stopped bringing ideas. Not dramatically. They simply stopped being the person who said "what if we tried this" in a meeting and became the person who answered the question they were asked. And then the work itself changed character. It was still good. It was still on time. But it had gone from creative to reactive. They were responding to the brief rather than improving it.

Every single deliverable landed. Nothing broke. And still I had quietly damaged one of the best people on the team.

The mistake was not adding the project. Sometimes you have to add the project. The mistake was the assumption underneath it: that high performers have unlimited capacity because they do not say no.

Key Takeaways

• Overload removes capability, it does not announce itself. Ideas go first, delivery goes last.
• Delivery is the last thing to break, and the only thing on the dashboard. By the time output slips, you have already lost the good part.
• Reliability is not availability. Your most dependable people are the least likely to tell you they are full.
• Delay is nonlinear. Past roughly 80% utilisation, wait time explodes rather than creeps.
• Plan for sustainable performance, not maximum utilisation. They look identical over a quarter and opposite over a year.

Overload Takes Things Away

Here is the thing I wish someone had told me earlier, because I think it is why so many good managers miss this. Overload does not add anything to a person's behaviour. It removes things.

We expect the opposite. We expect stress to announce itself. We are waiting for someone to look tired, to complain, to miss a deadline, to send a message at midnight that reveals the tension. Those are additive signals. Something appears that was not there before and you notice it. That is not how it goes.

In my experience, overload removes capability in a consistent order, and the order is exactly wrong from a manager's point of view. Call it the Overload Order.

First, the ideas go. Original thinking is the most expensive thing a brain does and the first thing cut when there is no room. The person stops suggesting improvements. They stop questioning the brief. Then proactive communication goes. They still answer everything, they just stop initiating. They stop flagging the thing that might become a problem in two weeks, because flagging it means owning it and there is no space to own anything else. Then creativity in the work itself goes. The output becomes competent and literal. It does what was asked and no more. Then, and only then, delivery starts to slip.

Look at that sequence and you see the trap. Delivery is the last thing to break, and delivery is the only thing on most dashboards. By the time your reporting tells you a person is overloaded, you have already lost three layers of what made them valuable, and you have probably lost them for a while, because those layers do not come back the moment the workload drops.

This is why judging capacity by output is almost useless for your best people. Their output is the last indicator to change. The real signal is what people stop doing, and you only notice it if you remember what they used to do.

Which brings me to something uncomfortable. Noticing this requires you to know what a person is like at their best. Not their metrics. Them. How they normally talk in a meeting, how often they usually push back, what kind of thing they usually notice. Without that baseline you cannot detect a decline from it, and you will keep managing on the only number you have.

The Overload Order: ideas go first, proactive communication next, then creativity, delivery last

Reliability Is Not Availability

The second part of my mistake was reading silence as capacity.

It is worth being honest about how this works, because it is not really about one person's personality. It is structural. You are asking someone whose professional identity is built on doing their best, in an industry where everyone is visibly busy, to volunteer that they have reached a limit. And you are asking at the moment when saying yes is easy and saying no requires them to argue against their own reputation. That is not a fair test. It is not even a test. It is a formality that produces the answer you were already expecting.

The people most likely to accept work they cannot sustainably carry are, almost by definition, your most reliable people. Reliability and self-protection are close to opposites. The quality that makes someone valuable is often the reason they do not tell you they are overloaded.

Which leads to a rule I now hold firmly: finding the limit is the manager's job, not the team member's. You cannot put that responsibility on the person you are giving more work to. If the only way you learn someone is overloaded is when they tell you, your system is not detecting overload. It is detecting who is comfortable speaking up, and that is a completely different thing.

The Maths Nobody Applies to People

There is a well-known principle in operations that explains how delay builds in any system. As a system approaches full capacity, delay does not rise gradually. It rises dramatically. Kingman's formula, the standard queueing result for wait time, puts the effect in a single term, utilisation divided by one minus utilisation, which is why you should never run a shared resource at 100%.

Here is roughly what that looks like as relative delay.

The pattern is not linear. Going from 80 to 90% does not add a little more waiting, it roughly doubles it. Going from 90 to 95 does the same again. Those last few points of "efficiency" cost an enormous amount, which is why the most effective teams do not try to keep people busy 100% of the time. They leave room for unexpected work, interruptions, and better decisions.

One important caveat. This is not an exact rule for people. Humans are not machines. We have motivation, experience, and the ability to change priorities. You cannot calculate the perfect workload for a marketing team with a formula. But the overall pattern is real, and anyone who has managed a team has seen it. A person at 95% takes longer to reply because every new request waits behind something else. Decisions slow because attention is already committed. Unexpected work fits nowhere, so every surprise becomes a problem.

The biggest difference from a factory is that you cannot see the queue. There is no pile of unfinished work on the floor. The queue is hidden. It is the message that has not been answered, the decision that has not been made, the task that has not been started. You only notice it when someone asks.

This leads to a conclusion many managers do not like: unused capacity is not waste. It is what allows a team to absorb the unexpected. And agency work is full of the unexpected. Clients change priorities. New requests appear without warning. Urgent issues arrive on a Friday afternoon. If every person is already at full capacity, there is nowhere for surprises to go, and every one becomes a late night, a missed deadline, or an escalation, not because the team is not good enough, but because there is no room left to absorb it.

The surprising part: teams with a little breathing room often deliver faster and more consistently than teams where everyone is busy all the time. Not because they work less, but because they always have room to handle what is next.

The utilisation curve: past 80%, delay rises far faster than load

The Queue You Cannot See

It is worth sitting with the hidden queue for a moment, because it is what makes overload so easy to miss until it is expensive.

In a factory, work in progress is physical. You can walk the floor and count it. On a knowledge team, the same backlog exists; it is just invisible. It lives as the reply that has not been sent, the brief that has not been reviewed, the half-formed idea that never made it to the meeting. Nobody can see it, so nobody prices it, so it grows until a deadline runs into it.

The job is to make that queue visible before it makes itself visible. That means asking questions that surface it rather than questions that flatter it. "How is your workload?" gets you nothing. "What did you not get to this week that you wanted to?" gets you something real.

The hidden queue: a factory backlog is visible, a team's backlog is not

From Maximum Utilisation to Maximum Sustainable Performance

So the way we plan resources now is different. We still look at availability, of course. But we spend roughly as much time on three other things: what someone's actual mental bandwidth looks like right now, where their motivation is, and whether the next three months move them forward or just keep them busy.

The biggest shift came when we changed how we defined success. The goal is not to maximise utilisation. It is to maximise sustainable performance.

Those two things look almost identical over a quarter and completely different over a year. Utilisation optimises the current sprint. Sustainable performance optimises the period over which a person keeps being excellent. Choose the first consistently and you get very good numbers for two quarters, then a resignation, or worse, a quiet version of a resignation where the person stays and the excellence leaves.

The practical translation is a question I now ask before assigning anything to someone already fully loaded. It is not "can they handle one more project." It is: what will they stop doing if I add this?

Because something always stops. That is not pessimism; it is arithmetic. The only question is whether you choose what stops, or whether the person chooses under pressure, and under pressure, people cut the invisible things first. Thinking time. Mentoring the junior person. The process improvement they were halfway through. The idea they were going to bring to the next meeting.

People do not create value only by completing tasks. They create it through judgment, curiosity, ownership, and commitment. Those are the qualities that improve a business month after month. If your resource planning protects only execution, you will always get execution. If it protects people's ability to think, improve, and contribute, you get something far more valuable: a company that keeps getting better because its people never stop making it better.

Utilisation versus sustainable performance: identical over a quarter, opposite over a year

Planning capacity for a growth program is most of what keeps quality steady as a team scales. It is the same discipline we describe in scaling delivery without losing quality.

Hours Are the Wrong Variable

There is a version of this argument that becomes soft and unhelpful, so here is a correction to my own position. The problem is not simply "too many hours."

Burnout is not only about how much people work. It is also about how they experience the work. People who feel motivated, supported, and connected to what they are doing can often handle long hours better than people who work the same hours but feel disconnected or unsupported. That cuts both ways. You cannot prevent burnout just by giving everyone less work, and you cannot assume a reasonable workload is enough if people do not feel supported or treated fairly.

Gallup's research on the five main causes of burnout makes the point cleanly. They are unfair treatment, an unmanageable workload, unclear communication from managers, a lack of manager support, and unreasonable time pressure. Only one of the five is about the amount of work. The other four are about how the work is organised, communicated, and led. In the same research, people who strongly agree they are often treated unfairly are 2.3 times more likely to burn out.

Unclear role expectations in particular should sound familiar to anyone who has thought about ownership. When people do not know what good performance looks like, or who owns what, they default to trying to do everything, and doing everything is the fastest route to exhaustion I know. Ambiguity makes people carry weight that was never theirs, permanently, because they cannot tell where their responsibility ends. This is exactly the tax described in our piece on team ownership. Clarity is not just an efficiency tool. It is a load-bearing part of how people stay well at work.

It also explains why two teams with identical workloads can have completely different outcomes. One knows exactly what it owns, gets consistent support, and understands why the pressure exists. The other is guessing. The hours are the same. The experience is not remotely the same.

Five causes of burnout, and only one of them is the amount of work

What Managing People Actually Involves

So if the number is not the thing, what is? The honest answer is that it is a relationship, and relationships do not compress well into frameworks. But there are things we try to do consistently.

1. Learn the individual patterns. People are not variations on one template. Some do their best work when you hand them a problem and disappear. Some need to talk it through before they can start. Some are motivated by scope, some by mastery, some by not letting the team down. Manage everyone the way you would want to be managed and you will do a good job only for the small number of people who are like you.
2. Know the strengths and the weaknesses, honestly. Not in a performance-review sense, in a working sense. Knowing that someone is brilliant at structure and weak at open-ended creativity changes what you hand them and how you brief it.
3. Make sure there is a path. Not a promotion promise, a direction. People tolerate a great deal of difficulty if they can see it is going somewhere, and almost no amount of comfort compensates for feeling stationary. This is one of the most common reasons good people leave and very rarely the reason they give.
4. Make the work interesting where you can. Not every task can be interesting, and pretending otherwise is insulting. But over a quarter, everyone should get something they actually want to work on. Interesting work is a retention mechanism and a quality mechanism at the same time.
5. Solve problems together rather than receiving them. There is a big difference between a manager who is handed a problem and issues a resolution and one who sits in it with the person for twenty minutes. The first is faster. The second teaches something and often produces a better answer, because the person closer to the work usually knows more about it than you do.
6. Give ownership and be clear about the limits. These go together and almost everyone does one without the other. Ownership without limits is not empowerment, it is abandonment, and the person finds the boundaries by crossing one and being corrected in public. Limits without ownership is micromanagement with extra steps. What people need is a real decision space, clearly bounded, with the boundaries stated up front rather than discovered.
7. Be a partner, not an inspector. Whether someone feels their manager is on their side or watching them changes everything about what they tell you. And what they tell you is your entire information system.

There is a lot of research behind this, and the idea is simple. Self-determination theory, from the psychologists Deci and Ryan, finds people are most durably motivated when three needs are met: some control over how they work, a sense of being capable, and a feeling of connection to the people around them. When those are met, motivation comes naturally and lasts. When one is missing, people start working only for external reasons, deadlines or money, or they gradually lose motivation altogether.

Two things matter here. Autonomy does not mean people do whatever they want. It means they understand why they are doing something and feel part of the decision, even when the task was assigned. That is why explaining the reason behind a decision matters. People take ownership when they understand the purpose, not just the instruction. And the three needs do not substitute for each other. If someone does not feel capable, more freedom will not fix it. If someone feels disconnected, more interesting work will not fix it. Managers often solve the wrong one, and it rarely works.

One finding stands out. Gallup attributes at least 70% of the variance in team engagement to the manager. That is a difficult truth if you are a manager, and an important one. If you want a company to grow, investing in great management is not optional. It is one of the highest-impact decisions you can make.

Heads of Department Are a Different Job

The way we work with Heads of Department is not a more senior version of how we work with individual contributors. It is a different activity. The shortest way I can say it: we do not manage their tasks, we manage the level at which they think.

With individual team members, most conversations are about the work itself, and that is exactly right. Priorities, blockers, feedback, progress. The point of those conversations is to make everything clear so the person can focus on doing great work instead of guessing what is expected. With Heads of Department, we almost never talk about individual tasks. If we are discussing a specific deliverable, something has usually gone wrong a level above it. Our conversations are about systems, decisions, trade-offs, hiring, the health of their team, and where they are spending their attention.

Attention is the real scarce resource at that level. A Head of Department can be extremely busy and still be pointed at the wrong problem, and being busy is very effective camouflage for that. So a question we ask often is some version of: what have you not looked at in a month that you probably should have? The answer is frequently the most important thing in the meeting.

The other expectation we hold is that Heads of Department bring recommendations, not just issues. If a problem lands, we want to hear what they think we should do. That single change moves a conversation from reporting to decision-making and develops judgment in a way nothing else does.

Now the correction, because this rule is dangerous without one. "Bring me solutions, not problems" sounds strong and can quietly cause enormous damage. Taken literally, it means bad news only reaches you once someone has an answer for it, which means the worst problems, the ones nobody has an answer for, reach you last, sometimes far too late. So our rule has an explicit exception, and we say it out loud rather than assuming people know it: bring the recommendation when you have one, and bring the problem immediately when you do not and it matters. Never sit on something serious because you have not solved it yet.

I would rather be handed a problem at 9am with no solution than a solution at 6pm to something that has been burning since Tuesday. The point of the recommendation habit is to build thinking, not to filter information. The moment it starts filtering information, it has become the opposite of what it was for.

The Multiplier Question

What we are ultimately trying to do with the leadership layer is make people multipliers rather than only problem-solvers. A problem-solver handles what reaches them and handles it well. A multiplier changes what their team is capable of. The difference is not talent, it is where the person spends their attention: on the queue in front of them, or on the system producing the queue.

Most people promoted into leadership were promoted for being outstanding problem-solvers. That is the skill that got them noticed and the hardest one to put down, because solving things feels good and building systems feels slow. Everyone in a leadership seat has felt the pull of just doing it themselves. It is faster today and a disaster over six months.

Which gives me the measure I actually use for my own job, and it is not a comfortable one. My success is not measured by how many decisions I make. It is measured by how many good decisions the leadership team can make without me. If more decisions are routing through me this quarter than last, that is not evidence I am important. It is evidence something is broken: either ownership is unclear, or people do not have the context to decide, or they have learned that deciding without me is risky. All three are mine to fix, and none get fixed by me making more decisions faster.

Hiring the person who runs a team well is the highest-leverage decision in an operation. If you are building one and want a partner who thinks about delivery this way, talk to us.

Where This Approach Breaks

4 Honest failure modes, because everything above can go wrong in the other direction.

1. Empathy can slide into avoidance. This is the big one, and I have watched good managers fall into it. Caring about someone's bandwidth is not a reason to postpone a conversation about their performance. If someone is not delivering, understanding why is essential and excusing it indefinitely is a disservice to them and unfair to everyone around them. High support and high standards belong together. High support with drifting standards is not a kind culture, it is a slow one, and the strongest people leave it first because they can see exactly what is happening.
2. Protecting people from load is its own harm. Growth requires stretch, and stretch is uncomfortable by definition. Read every sign of strain as a reason to cut someone's scope and you will keep them comfortable and stall their career. The line between a stretch that builds someone and a load that damages them is not a formula. It depends on the person, the moment, and the support around it. I get it wrong in both directions, and the only thing that helps is being close enough to correct quickly.
3. Reading bandwidth remotely is genuinely hard. We are over 60 people across seven departments and most of the world's time zones. Nobody walks past anybody's desk. You lose the signals a shared room gives you for free, the tone, the posture, the fact that someone has gone quiet in a way that is not like them. What replaces it has to be deliberate: asking directly, asking more than once, and understanding that the first answer is very often "I'm fine."
4. One person cannot hold sixty people's context. We cannot know everyone's motivation, patterns, and growth path at the depth this requires. Nobody can. Which is the real reason the Head of Department layer matters more than any other hiring decision in the company. Promoting or hiring the wrong person into a manager role is not one bad decision. It is one bad decision multiplied by everyone who reports to them, compounding quietly.

Headcount Is a Number You Report

The word "headcount" is useful for exactly one thing: telling somebody how many people are in a company. It is a reporting unit, and it is fine.

The trouble starts when it becomes a planning unit, when the question in your head shifts from "who does this well and what does it cost them" to "do we have the bodies?" Because 60 people is not a capacity. It is 60 different capacities, each of which changes depending on what else that person is carrying, how clear their role is, whether they can see where they are going, and whether the person they report to is any good.

Stop planning for how much work people can take on. Start planning for how much value you want them to keep creating. Good resource planning is about creating the conditions that let great people keep doing great work for a long time.

Frequently Asked Questions (FAQs)

Final Word

If I could compress all of this into a single instruction, it would be to stop planning for how much work people can take on and start planning for how much value you want them to keep creating. The first number is easy to see and quietly destructive. The second is harder to measure and the only one that compounds.

Related Reading

• Fixing "I Thought You Owned That". Why role clarity is a load-bearing part of how people stay well at work.
• Scaling Marketing Delivery Without Losing Quality. Capacity planning as a team grows past quick conversations.
• How to Hire a Crypto Marketing Team. Why the manager hire is the one that compounds.

Sources

• Kingman's formula and queueing theory, on utilisation and wait time link: en.wikipedia.org/wiki/Kingman's_formula
• Gallup, employee burnout part 1, the five main causes link: gallup.com/workplace/237059/employee-burnout-part-main-causes.aspx
• Gallup, managers account for 70% of variance in team engagement link: news.gallup.com/businessjournal/182792/managers-account-variance-employee-engagement.aspx
• Deci and Ryan, self-determination theory (autonomy, competence, relatedness) link: selfdeterminationtheory.org/theory

Disclaimer

This article reflects operating experience at AP Collective and is general management guidance, not a guarantee of results. Queueing figures are illustrative of a nonlinear pattern, not an exact model of human teams. Cited studies are attributed to their original researchers and publishers.

Written by Lui, Chief Operating Officer at AP Collective.