How to Design Roles Around Decisions, Not People

Most senior roles in startups are built around a person. Someone good needs more scope, or a strong candidate is available and you assemble a job that fits them, or a long-serving employee has grown and you formalise whatever they have accumulated. It is a reasonable way to keep good people and it produces a specific problem: a role only that person can do, whose boundaries nobody can articulate, and which becomes unfillable the moment they leave. This page is the alternative — scoping roles by the decisions they own, and testing them before anyone is in the seat.
What’s inside
Person-shaped roles are the default, and the problem
What a decision-shaped role looks like
The four-part role definition
How to test a role before you hire into it
When person-shaped is actually correct
Fixing an org that is already person-shaped
Where Blomma fits
Person-shaped roles are the default, and the problem
A person-shaped role is one whose scope is the union of what a particular individual is good at and what happened to be unowned when they arrived. It looks like: Head of Operations who also owns customer support, the data team, and vendor contracts, because those are the four things the person who took the job was willing to pick up.
That role works well while that person holds it. Three problems arrive later.
It is unfillable.When they leave, the job description is a list of unrelated things and any candidate is a poor fit for some of it. You either split the role in a hurry — during a transition, which is the worst time — or hire someone who fails at a quarter of it.
It hides gaps.A role assembled from whatever was unowned means nobody ever asked what the company actually needs owned. Things that matter and were not in anyone’s range remain unowned indefinitely.
It makes performance unassessable.With no defined outcome, evaluating the person is a judgment about the person rather than a comparison against a standard. That is how founders end up unable to say whether a senior leader is doing well, and unable to have a useful conversation about it.
The fix is not to stop accommodating good people. It is to define the role first, then decide who fits.
What a decision-shaped role looks like
A decision-shaped role is defined by the outcome it owns and the decisions it holds to deliver that outcome. Everything else — activity, team size, title — follows.
Compare. Person-shaped: “VP Operations — owns ops, support, data, and vendor relationships.” Decision-shaped: “Owns cost-to-serve, measured by gross margin per account. Decides: support staffing model, tooling spend to fifty thousand, vendor selection, service-level targets. Escalates: anything changing the customer-facing commitment, spend above fifty thousand.”
The second version tells you what the role is for, what success looks like, whether the person has the authority to deliver it, and what a candidate needs to have done before. It also makes it immediately visible if the role has been given an outcome without the decisions required to affect it — which is the single most common defect in startup senior roles and the reason capable senior hires fail.
That defect is worth dwelling on. If someone owns an outcome but three of the five decisions that drive it sit elsewhere, they will fail regardless of ability, and the failure will be attributed to them.
The four-part role definition
One short document per senior role. Fifteen minutes each, and it prevents most of what goes wrong.
Outcome.One sentence, ideally with a metric. What is different in the company because this role exists? If you cannot write this, you are not ready to hire.
Decisions owned.Five to eight specific decisions this role makes without escalating. Be concrete — “pricing for the mid-market segment,” not “commercial strategy.” This is the part that converts a title into authority.
Boundaries.What escalates, and at what threshold. Spend limits, decisions touching other functions, anything with legal or customer-commitment implications. Explicit boundaries are what make real autonomy inside them possible.
Measure and cadence.How you will both know whether this is working, and when you will look. Without this, the first performance conversation is an argument about impressions.
Then check one thing before you finish: does the person holding this role have the decisions needed to move the outcome? If not, either move the decisions or change the outcome. Do not proceed with a mismatch and hope seniority resolves it.
How to test a role before you hire into it
Four tests, all cheap, run before a search starts.
The candidate-profile test.Write the profile from the role definition. If the resulting profile is internally contradictory — needs deep technical judgment and to be a strong enterprise closer — the role is two roles. Better to find that now than after two failed searches.
The decision-trace test.Take the five decisions the role owns and check who makes them today. If three of them currently sit with you and you are not genuinely willing to hand them over, the role as written is fiction.
The peer test.Name who this role has to negotiate with to succeed. If it must reach agreement with three peers on its core decisions, the role has responsibility without authority and no hire will fix it.
The absence test.If this role were vacant for three months, what specifically would deteriorate? A clear answer means the role is real. A vague one means you may be hiring for capacity or for status rather than for an outcome.
When person-shaped is actually correct
Worth being fair, because the pure version of this advice is too rigid for early companies.
Before roughly thirty people, person-shaped is often right. You need range over specialisation, the work changes monthly, and a role defined tightly enough to be decision-shaped would be obsolete in a quarter. Formalising too early costs flexibility you cannot afford.
It is also correct when you have an exceptional person and a genuine choice between building a role around them or losing them. That is a real trade and sometimes the right call — with one condition: write down that you have done it, and what the role would look like rebuilt from scratch. That note is what saves you when they leave.
And it is correct for genuinely novel work where nobody knows yet what the decisions are. Define the outcome, leave the decisions loose, and revisit in six months once the shape is clear.
The failure is not person-shaped roles. It is person-shaped roles that nobody ever converts, accumulating until the org chart is a record of who was available.
Fixing an org that is already person-shaped
Most companies past fifty people have several. Converting them does not require a reorg.
Start with the roles that matter most and are least well defined — usually one or two senior seats where you could not say what success looks like. Write the four-part definition for what the roleshouldbe, from the company’s needs rather than the current occupant’s strengths.
Then compare it with what the person actually does. Three outcomes: it mostly matches, and you have just given them clarity they lacked; it matches partly, and you have found scope to move; or it barely matches, in which case you have discovered something important and should not act on it in a hurry.
Do this as a scheduled annual exercise rather than in response to a problem. A role redefined during a scheduled review is planning. The same redefinition proposed after a bad quarter is a judgment, and it will be received as one.
And do it before you hire the next senior person, because a new decision-shaped role sitting next to three person-shaped ones will produce boundary conflicts immediately.
Where Blomma fits
The uncomfortable part of role design is that doing it honestly surfaces two things founders avoid: that some of your senior people hold scope that does not match what the company needs, and that some of the decisions you have nominally delegated still sit with you.
Blomma is an always-on AI career coach with no stake in your structure. Use it to write the four-part definition for a role you are about to create, and to run the decision-trace test honestly — including whether you are genuinely willing to hand over the decisions the role requires. Use it to check whether a role you are about to hire for is actually two roles. And use it to prepare the conversation when a redefinition means someone’s scope changes, which is the part of this work that determines whether it lands as clarity or as demotion. Where the situation warrants a human who has scoped senior roles at your stage, bring one in..
A role is a set of decisions with an outcome attached. Define it that way and you can hire for it, assess it, and fill it again when someone leaves. Define it around a person and you have created a dependency you will discover the cost of at the worst possible moment.
