How Team Size and Roles Change at Each Startup Stage

A startup's team doesn't just get bigger as it grows. It changes shape. Founders who once did everything hand work to specialists, specialists get managers, and managers eventually need managers of their own. Each shift changes how decisions get made, who talks to whom, and what the founders themselves are supposed to be doing.

There's no official headcount table for "seed" or "Series A," and this article won't invent one. Funding stages measure money raised and milestones reached, not people. What we can do is lay out the named models that researchers and practitioners use to describe how organizations change with size: Greiner's growth phases, Dunbar's number and its limits, span of control research, Churchill and Lewis's small business stages, and the family-tribe-village-city-nation ladder from blitzscaling. Where a source gives a number, we cite it. Where it doesn't, we describe the pattern in words.

Why headcount alone is a poor guide

Two startups with 30 employees can be organized in completely different ways. One may have three founders still making most decisions. The other may already have functional heads, a hiring process and a formal budget. Both are "30 people," and they face different problems.

That's one reason the best-known models describe structure rather than headcount. Churchill and Lewis, in their well-known 1983 paper, criticized earlier life cycle models for defining size mostly by annual sales and overlooking other factors, according to a review in the Journal of Organization Design. The factors they named included value added, number of locations, and the complexity of the product line. Headcount is a rough proxy for all of this, nothing more.

So read the stages below as patterns of work and coordination. The numbers that do appear come from specific frameworks, and each is labeled as such.

Stage 1: Idea and pre-seed, founders only

At the very start, the team is the founding team. Nobody has a department. Founders choose what to build, talk to prospective customers, write the first version, and handle money, legal setup and hiring by default because nobody else exists to do it.

Two things are true of this stage regardless of the exact headcount:

• Roles are fluid. A founder with a technical background may also be the first salesperson. A founder with a business background may be doing support. Titles exist mostly for outsiders.
• Coordination is free. With a handful of people, everyone knows what everyone else is working on. Almost nothing needs to be written down.

The blitzscaling framework gives this ordering a number. Chris Yeh, co-author with Reid Hoffman, describes the first stage as a "family" of 1 to 9 people, as he explains in a ServiceRocket podcast interview. Churchill and Lewis describe something similar in their first stage, Existence. In a summary of the model, the owner handles all responsibilities, directly supervises a small number of workers, and no formal systems or planning exist.

For how founders divide the work among themselves, see cofounder roles and what a founding team is.

Stage 2: Seed, the first hires

The first employees usually arrive around the seed stage. What matters about these hires isn't how many there are but what kind of person fills the role. Early hires tend to be generalists: people who can switch between building, selling and supporting, and who are comfortable without a defined job description.

Typical first-hire categories, described qualitatively:

• Builders. Engineers or designers who extend the product beyond what the founders could build alone.
• Sellers and customer-facing people. Someone who takes over customer conversations from a founder, so a founder can return to product or fundraising.
• Operators. A person who handles the administrative load that builds up: payroll, contracts, basic finance.

These people also hold a different position from later employees. They often receive equity from an employee option pool, and they influence the product and culture in ways a later hire in a defined role won't.

The org chart at this stage is still flat. Founders manage everyone directly. In Churchill and Lewis's terms, the company has moved from Existence into Survival, where, per the same summary, the organization stays simple with the owner still in control and the central question is whether revenue covers expenses.

Dunbar's number, and why not to lean on it

Anyone writing about team size eventually meets Dunbar's number, the claim that humans can maintain stable relationships with about 150 people. It's tempting to treat 150 as the point where a company must add structure.

It comes from an extrapolation. Robin Dunbar related primate neocortex size to group size and extended the line to humans. A 2021 paper in Biology Letters, "Dunbar's number deconstructed" by Patrik Lindenfors and colleagues at Stockholm University, reran the analysis with modern statistical methods. Stockholm University's summary of the study reports that the 95 percent confidence intervals for the estimates ran from 2 to 520 people, and it quotes the authors as saying it is not possible to make an estimate for humans with any precision using available methods and data.

So Dunbar's number is a useful reminder that personal familiarity doesn't scale forever. It isn't a measured threshold for companies. If someone tells you a startup "must" restructure at 150 employees because of Dunbar, they're repeating a figure its own critics say can't be pinned down. A team's real strain point depends on how work is divided and how information moves, not on a cognitive constant.

Stage 3: Series A, specialists replace generalists

By the time a company has some evidence the product works and is raising to scale it, the work has outgrown generalists. The same person can no longer own sales, onboarding and support. Roles begin to separate.

This is where the shape of the team changes most visibly:

• Functions form. Engineering, product, sales, marketing, customer success and operations start to look like separate groups, even if each has only a few people.
• Hiring shifts from "can do anything" to "has done this before." A company that needs its first dedicated sales leader or head of engineering is hiring for depth in a function it has not run before.
• Founders start managing managers. Instead of coordinating every person, a founder works with a small number of leads.

Blitzscaling's "tribe" stage, 10 to 99 people per the Yeh interview, is the range where an informal structure gets stretched, which is why it overlaps the seed and Series A years for many companies. Yeh calls the next jump, from tribe to village, the hardest transition, because families and tribes run without formal processes, while a village (100 to 999) needs real formal structure. He pegs the change at around the 100-person mark.

Greiner's model describes the same pressure in management terms. His first phase, growth through creativity, ends in a crisis of leadership because informal communication stops working as the team grows, according to Toolshero's overview. The fix is usually a more formal structure and professional management, which is the second phase, growth through direction. Our full explanation of Greiner's growth model covers all the phases. For how the founder's role itself shifts during this, see founder CEO skills by stage.

How many people can one manager handle

Once management layers appear, the question becomes how many direct reports each manager should have. This is span of control, and the evidence is less tidy than the popular rules of thumb.

Gallup's research is one of the larger recent data sets. According to Gallup's span of control report, the average number of people reporting to managers rose from 10.9 in 2024 to 12.1 in 2025. The median team size has held at about five to six employees per manager or leader. Gallup also reports that 37 percent of managers oversee fewer than five people, 66 percent manage fewer than ten, 22 percent have 10 to 24 direct reports, and 13 percent oversee 25 or more. Its underlying analysis covered 92,252 teams across 104 organizations and 26 industries, in employees from 46 countries.

Two things stand out for a startup. First, the typical manager leads a small team, but a minority of very large teams pulls the average up. Second, Gallup found that managers spend a median of 40 percent of their time on individual contributor work, so a startup manager who is also a hands-on contributor isn't unusual.

The practical point: there isn't one correct span. Narrow spans mean more managers and more coordination cost. Wide spans mean less attention per person. A startup tends to start wide (a founder with many direct reports, because there are few layers) and narrows as managers are hired.

Churchill and Lewis: when growth is a choice

Neil Churchill and Virginia Lewis published "The Five Stages of Small-Business Growth" in Harvard Business Review in May 1983. The stages, as described in the Saylor summary, are Existence, Survival, Success, Take-off and Resource Maturity.

The model matters for team structure for two reasons:

1. It treats stability as legitimate. At the Success stage the owner can choose either to grow, with functional management, or to stay stable and disengage, with managers taking over operations. The Journal of Organization Design review notes that Churchill and Lewis criticized older models for assuming a company must grow through every stage or die.
2. It ties growth to delegation. At Take-off the organization becomes decentralized and the owner must decide how to delegate and how to finance rapid growth. At Resource Maturity, the same summary describes decentralized management with experienced staff, comprehensive systems, and separation between owner and business.

Not every company wants to become a large one. If the plan is a steady, profitable business of modest size, the Success stage is a destination, not a stop on the way. A venture-backed startup is a different case, since its financing assumes growth, but the model is a useful check on whether a given team size is a decision or an accident.

Stage 4: Growth, management layers appear

After Series A, a successful company often grows fast. The pattern shifts from "add specialists" to "add structure."

• Departments become teams, and teams get managers. A head of sales now has regional leads, a head of engineering has team leads.
• A leadership team forms. Functional heads meet as a group, and the founder's job is increasingly setting direction and hiring the people who run things. See the management team below the founder.
• Processes get written down. Hiring, onboarding, budgeting and planning stop living in people's heads.
• Sales organizes. Revenue teams split into roles such as prospecting, closing and account management.

Greiner's phases two and three describe the tension here. Direction from the top brings order but frustrates people closest to customers, which produces a crisis of autonomy. Delegation fixes that and can lead to a crisis of control, where top managers feel they're losing their grip, per Toolshero.

Stage 5: Scale-up, coordination and specialization at depth

At the scale-up end, the company has multiple layers and many functions. Headcount in the hundreds or thousands is common in the blitzscaling ladder: village is 100 to 999, city is 1,000 to 9,999, and nation is 10,000 or more (Yeh interview). Most startups never reach the last of these.

What changes in the work:

• Roles get narrower. People specialize within a function: a sales team has sales operations, enablement and separate segments.
• Coordination becomes a job. Planning, finance and program management exist because the number of connections between teams is too large to manage informally.
• Culture has to be maintained on purpose. What a small team gets from sitting together, a large company must build through values, rituals and clear communication.

Greiner's later phases, growth through coordination and then collaboration, address this territory. The risk named for phase four is red tape: systems that outlive their usefulness, with procedure taking precedence over problem solving. Collaboration is meant to answer it with teams and looser control. For what defines that stage, see what a scale-up is, and for the speed-first view of scaling, blitzscaling.

Summary: stage, team shape and model

The table pairs each stage with the qualitative shape of the team and the model labels that fit. Headcount appears only where a named source gives a range.

Treat the table as a map of labels, not a prediction. A company can sit in the "village" range with a founder-led culture, or have 20 people and a formal structure.

Key Facts

• Chris Yeh describes the blitzscaling stages as family (1 to 9), tribe (10 to 99), village (100 to 999), city (1,000 to 9,999) and nation (10,000 or more), and calls tribe to village the hardest transition (ServiceRocket).
• Lindenfors and colleagues found 95 percent confidence intervals of 2 to 520 people around estimates of a human group-size limit, so the 150 figure can't be pinned down (Stockholm University).
• Gallup reports average span of control rising from 10.9 in 2024 to 12.1 in 2025, with a median team size of about five to six (Gallup).
• Churchill and Lewis published their five-stage small business model in HBR in May 1983 (HBR).
• Greiner described five growth phases, each ending in a crisis, and later added a sixth (Toolshero).
• No source cited here gives an official headcount for seed or Series A. Funding stage and team size are separate things.