How to Structure Your Team as You Scale
There is no correct structure for a startup. There is a correct structure for a startup at a particular size, doing a particular kind of work, and it stops being correct roughly every time you double. The practical problem is that nobody announces the transition — the structure that worked keeps working slightly less well, and by the time the friction is obvious you are usually one stage behind. This page maps the four stages, what each one requires, what breaks at the end of it, and how to see the next transition coming rather than discovering it.
What’s inside
Structure is stage-specific, not a preference
Stage one: under 15
Stage two: 15 to 50
Stage three: 50 to 150
Stage four: 150 and beyond
The transition signals
What to build one stage early
Where Blomma fits
Structure is stage-specific, not a preference
Most debates about startup structure are really debates about stage. Someone advocating for cross-functional pods and someone advocating for functional depth are frequently both right, about companies of different sizes.
The reason is coordination cost. At eight people, everyone holds the whole context and coordination is free, so any structure works and the lightest one wins. At eighty, nobody holds the whole context, and the structure determines which conversations happen easily and which require a meeting with six people in it. Structure is how you decide where to spend coordination cost.
Which means the question is never what structure is best. It is what your current coordination cost is, where it is being paid, and which structure moves it somewhere cheaper. That question has a different answer at every stage, and the answer changes before you notice.
One consequence worth stating early: you will run every structure you ever adopt slightly too long. That is normal, because restructuring is expensive and the signals are ambiguous until they are obvious. The goal is not to be perfectly timed; it is to be one stage behind rather than two.
Stage one: under 15
What it looks like:everyone reports to a founder, no managers, no functional boundaries that anyone enforces. People do whatever is needed.
Why it works:coordination is free because everyone hears everything. Specialisation would cost more than it returns — you need range, not depth.
What to build:almost nothing structural. What you do need is written context, because the thing that scales worst from this stage is the shared understanding that currently lives in conversation. Start writing decisions down now; it costs an hour a week and saves months later.
What breaks:the founders’ calendars. Every decision routes through two or three people, and at around a dozen reports each that stops working. The first symptom is usually not chaos — it is a founder who is permanently behind and a team that is waiting.
Stage two: 15 to 50
What it looks like:functional teams with a lead each — engineering, product, sales, marketing, ops. One management layer. Founders manage the leads.
Why it works:functional structure gives you craft depth and consistent standards at the point where quality starts varying between people. It also gives you an obvious place to put each new hire, which matters more than it sounds when you are hiring monthly.
What to build:the first managers, and — critically — support for them, because most will be first-time managers promoted from doing the work. Also written decision rights, because this is the stage where “who decides” starts producing real friction. And a weekly forum where the functional leads resolve cross-functional issues without a founder present.
What breaks:the cross-functional seam. Anything requiring three functions to agree becomes slow, and the founder becomes the integration point by default. This is the most common place for companies to stall, and it usually gets diagnosed as an alignment problem rather than a structural one.
Stage three: 50 to 150
What it looks like:either functional groups with managers underneath the leads, or a shift to cross-functional pods owning outcomes, or some combination. Two management layers. An executive team starts to exist.
Why the fork:this is where the functional-versus-pod decision genuinely bites. If you have one product and craft quality is your differentiator, deepen the functional structure. If you have several distinct customer problems that can be worked in parallel, pods will move faster. Getting this wrong is recoverable; not deciding it is worse than either choice.
What to build:an executive team that resolves things among themselves rather than escalating to you — about a year of work and the highest-leverage thing available at this stage. Managers of managers, which is a genuinely different skill nobody has yet. And systems that make work visible without routing through you: shared boards, written status, metrics that anyone can see.
What breaks:you. Specifically, the founder’s ability to be the only person who can see across the whole company. Also the informal information flow — at this size, if a decision’s reasoning is not written down, three teams will act on three different understandings of it.
Stage four: 150 and beyond
What it looks like:semi-autonomous units — divisions, business lines, or large product groups — each with enough function inside it to operate. Three or more layers. Shared platforms and services underneath.
Why it works:autonomy is the only thing that scales past the point where any one person can hold the company. Each unit gets clean accountability and its own numbers.
What to build:genuine decision authority inside the units, or you have added layers without adding speed. Explicit interfaces between units and shared services. And a real answer to where development comes from, because at this size managerial attention per person is thin and coaching is the first thing a wide span destroys.
What breaks:duplication, drifting standards between units, and platform work that no unit wants to fund. Also culture — this is the stage where a company stops having one and starts having several, one per unit, unless someone actively holds the line.
The transition signals
More useful than headcount, because the thresholds above are approximate and your business may move faster or slower.
Time to a decision is rising.Track how long it takes for a typical cross-functional decision to get made. If it is lengthening, you have outgrown the structure regardless of headcount.
The same conflict escalates to you monthly.A recurring escalation is a structural gap with a name. One or two are fine. More than that and the structure is not resolving what it should.
New hires are not adding output.Headcount up, throughput flat. The classic signature of coordination cost exceeding the marginal contribution of another person.
Your best people are mostly coordinating.When strong contributors spend more time in alignment meetings than doing the work, the boundaries are drawn in the wrong place.
Nobody can name the owner of something important.Ask three people who owns retention. If you get three answers, the structure has a hole.
What to build one stage early
The useful asymmetry: structure should be built slightly late, and the things that support structure should be built early. Four that are cheap now and expensive later.
Written decisions and reasoning.Start at ten people. The context in founders’ heads is the single biggest thing that fails to scale, and writing it down is the highest-return unglamorous work available.
Manager support.Your first managers will be promoted contributors with no training. Build the support before you promote them, not after they struggle — the failure of a first-time manager is usually a failure to prepare them.
Decision rights.Write them at thirty people, not eighty. They read as bureaucratic early and they prevent the friction that arrives on schedule.
A forum where leads resolve things without you.Establish it while the issues are small, so the habit exists before the stakes are high.
Where Blomma fits
Two applications, and they are at different altitudes.
For you: the structural calls above are made with everyone around you positioned. Your leads have a view on whether their function should be split, your strongest hire has a view on whether pods would suit them, and none of them can advise you neutrally on a chart that determines their own scope. Blomma is an always-on AI career coach with no stake in your structure — a place to work out which stage you are actually in, whether the functional-or-pod call is being deferred rather than made, and how to sequence the individual conversations a change requires.
For the layer below: every structural decision you make sets how much managerial attention each person receives. The stage-two and stage-three transitions in particular create a wave of first-time managers and managers-of-managers who have never done the job, at exactly the moment your own capacity to develop them is thinnest. Coaching that does not depend on a manager’s calendar having room is how you avoid the standard outcome, which is a company that grew its headcount and not its bench..
Structure is not a thing you get right once. It is a sequence of stages, each of which works until it does not, and the skill is reading the transition before the friction becomes obvious to everyone else.