Accountability Chart: EOS seats, roles and who fits

An org chart records the structure you have, drawn from the people already in it. An Accountability Chart proposes the structure the company needs, drawn from the work, with names added last. Same shape on paper, but built from the other end. That reversal is what makes the second one uncomfortable to fill in, and it is where all of the value sits.

The practical difference shows up in what a box contains. An org chart box holds a person and a job title. The other holds a seat and the three to five things it is answerable for, each written as an outcome. Duties do not appear. Anyone reading the page can tell whether the seat is being filled well.

That sounds like a formatting preference until you draw it for a company that grew fast. Then two boxes turn out to hold the same accountability, one box holds nine, and a founder is quietly sitting in four of them. None of that is visible on a reporting diagram, because a reporting diagram was never asked the question.

Because putting names down first turns the exercise into a defense of the current arrangement. Once a leader sees their own box, every question about the structure reads as a question about them. Drawing the empty seats keeps the conversation on what the business needs for the next year, which is the only version of it a team can have honestly.

The usual method is to work from the top down with no names visible: what are the three or four functions this company must perform to hit next year's plan, what does each one own, and what sits under it. Twenty minutes in, somebody will say that a seat on the diagram does not exist yet. That is what the team came to find.

Adding the names afterward is a separate sitting, deliberately. By then the seats have been agreed on their merits, so the conversation about who fits stays on the person and the seat. Nobody has spent an hour arguing for a box because it happens to be theirs.

Two seats at the top, three under them. The Visionary owns direction, relationships and the ideas nobody asked for. The Integrator owns the day to day and holds the others to the plan. Underneath sit sales and marketing, operations, and finance. Five boxes, sometimes three in a company under twenty people, and that is a complete first draft.

A first draft usually mirrors the company already in front of you: fifteen boxes, one for every manager, because leaving somebody off feels like a demotion. Hold that off for an afternoon. Five boxes force the harder question: what must the business do to hit next year's number? The answer is never fifteen things.

Those three functional seats hold up across industries because the work does. Revenue has to come in. Sold work has to get delivered. The money has to stay straight. A software company calls the middle one engineering and a builder calls it operations, and the box does the same job in both. Add a fourth when a function is genuinely big enough to answer for its own results, not because a person deserves a box.

A short outcome the seat is answerable for, clear enough to say out loud and call met or unmet without a side debate. Sales pipeline kept full. Jobs delivered on the date quoted. Cash collected inside terms. Three to five of those per seat. Activities do not qualify, because every activity can be performed diligently in a quarter where the outcome never arrived.

Three to five per seat is the working range. Fewer than three usually means you have written down a task and called it a function. More than five and nobody can hold the list in their head, which defeats the purpose: the person in the seat and the person they answer to have to carry the same items.

A useful test is to read an accountability out and ask what number would tell you it was being met. An obvious answer means that number probably belongs on the weekly board. No answer at all means the line has been written as an activity, and it needs rewriting before anyone is measured against it.

The three seats below are ours, written the way a leadership team writes them on the day: outcomes in the company's own words, short enough to read aloud. Yours will use different nouns. What carries over is the shape, a result with a number somebody could point at.

The builder below is our own version, not the official worksheet. It starts with the three functional seats. Add more, title each one, write up to five roles as outcomes, put a name in or leave the seat open, and choose which seat it answers to. A name in two seats gets flagged. Copy the draft as indented text or CSV.

Fill in the titles and the roles before any names. That is the point of the exercise. Expect the argument to start at the fourth seat. If you cannot write three outcomes for a box, it is a task somebody does, not a seat, and it belongs inside another one. Take the finished draft into the quarterly session rather than treating it as the chart.

Every seat gets exactly one name, and a person may hold several seats. That rule is where the first real disagreement happens, because co-leadership of a function is common in 10-to-250-person companies and the diagram will not represent it. Two names on a seat means a decision has two owners, which in practice means it has none.

The founder usually holds three or four seats on the first draft, which is accurate and also the point. Seeing it drawn starts the conversation about which one gets handed over first. The honest answer is usually the seat they are worst at, which is seldom the one they enjoy least.

Where two people genuinely share a function, split the accountabilities and leave the box alone. One owns new business, the other owns delivery. If the accountabilities cannot be split, the two of them are doing one job and the company has a harder conversation waiting for it.

In this order: draw the seat, agree what it owns, then test a person against it. The test has three parts. Do they get it: the job makes sense without being explained twice. Do they still want it on a wet Tuesday in February? Can they carry it at the volume the seat now runs at? All three, or the answer is no.

Order is the whole trick. Test a person against a seat nobody has written down, and the room argues about the person, the conversation everybody was avoiding in the first place. Test them against five written lines and you get an argument about the lines. That one can be settled.

Which of the three fails tells you what happens next, and the answers are not interchangeable. Capacity often means hours, a tool, or one hire underneath. Of the three failures, it usually costs the least to fix. Getting it either shows up in the first few months or never does. Wanting it cannot be coached at all, and somebody who has quietly stopped wanting a seat will keep performing it competently for a year while the company slows down around them.

One person may hold two seats. Two people may never hold one. Below thirty people the first is ordinary and worth writing down, because it creates a scheduling problem you can see coming. The second leaves a decision with nobody answerable for it. That kind of problem gives no warning at all.

The asymmetry is not arbitrary. Somebody in two seats can be scheduled. They know which hat Thursday belongs to, and when the two collide, one person has to choose. A seat with two people in it has no such mechanism. The call goes to whoever cares more that week, and six months later nobody can say who was answerable for the result.

For everything administrative. Payroll, approvals, onboarding, holiday cover and every HR system in the building run on reporting lines and always will. The two diagrams are not competing for the same job, and a company that deletes its org chart in favor of the other one has swapped a useful record for a planning tool.

Most companies that use both keep the reporting diagram in the HR system, where it updates itself as people join and leave, and keep the seat diagram in the same place as the annual plan, where it gets reviewed when the plan is reviewed. Different documents, different update cycles, different questions.

The one thing to avoid is maintaining two diagrams that disagree about who reports to whom without anyone noticing. If the seat diagram says one thing and the payroll system says another, somebody is being managed by two people, and that is worth finding out deliberately, before a difficult quarter finds it for you.

At fifteen, expect five boxes, three names, and a founder who appears in two or three of them. At sixty it is closer to a dozen boxes, one name each, with a second row under sales and under operations. The seats that arrive in between are the ones nobody owned at fifteen: hiring, scheduling, and whoever answers for the numbers.

At fifteen people the page looks almost too simple to be worth drawing. Five seats, three names, one of those names in three boxes. Finance is a bookkeeper and the founder. Marketing is whoever has an hour. The value at this size is not the structure, which everyone in the room could already recite, but the lines written underneath each box. They are the first written statement of what the company expects of itself.

At sixty the same page has grown a second row and lost its ambiguity. Sales has separated from marketing, because the two answer for different weekly numbers. Operations carries two or three seats beneath it, usually delivery, scheduling and quality. Finance is a controller with a real close date. Somewhere in there a seat appears that owns people work: hiring, onboarding, reviews, the handbook. That work existed at fifteen too. It was spread across three people who each did a quarter of it badly.

Take an hour with the leadership team and a blank page. Write the functions the company has to perform, then the outcomes each one owns, then what sits beneath it. Leave the names off until a second sitting. Expect the first draft to be wrong in two places, and expect one of those two to be a seat nobody is in.

Draw it for the company you expect to be in twelve months. Not the one you are today, and not the one you hope to be in five years. A seat you will need next year is worth putting on the page empty. A seat for a division that does not exist yet is a distraction, and it will be redrawn before you get there.

The leadership team draws this page on the Focus Day®, the first working session of an EOS® implementation, before the vision document gets finished across the two Vision Building® days that come after it. Implementers describe it as the hardest of the early sessions, which is a fair expectation to walk in with. The discomfort is the tool doing its job.

Once a quarter the leadership team redraws the page from blank instead of editing it. A blank sheet takes twenty minutes and catches what an edit never will, because an edit defends whatever is already on the paper. Then every seat gets two questions: does this seat still match the work, and is the right person sitting in it?

Redrawing sounds wasteful. It costs less than the alternative. Give one person a blank page and ten minutes to sketch the seats next quarter's plan needs, then set the old version beside it. Those differences become the agenda. An edit yields two tidy changes. A redraw turns up the box somebody has been carrying that nobody ever named.

Keep the two questions apart, because their answers run in opposite directions. A seat that no longer matches the work gets rewritten, and nobody's job is on the table. A seat whose holder no longer fits is a conversation with a name attached, and it belongs in a private room. Teams that run the two together end up doing neither one honestly.