पाठशाला Pathshala · दल Dal, The team · Lesson 24 · Scale

Managers of managers: building the second layer

A company runs without its founder only when someone can hire, develop and direct the first-line managers. When that layer arrives, what it does, and how to fill it without promoting your best engineer by default.

Pathshala, The Founder Library · 11 October 2026 · 7 min read

An aerial view of golden rice terraces stepping down a hillside around a village. — Photograph: Quang Nguyen Vinh · Pexels

Most founders build the first layer of management by necessity: the team grows past what one person can direct and a few people become managers. The second layer, managers whose reports are themselves managers, is the one that lets a company run without its founder in every decision. It is also the layer founders most often fill by promoting whoever has been there longest or codes best, and then wonder why the company slows down.

This lesson works out when the second layer arrives, what the job is, who should do it, whether to promote or hire, and how to know if it is working.

When the second layer arrives

The arithmetic is set by span of control, the number of people one manager can lead well. GitLab’s handbook puts it at around seven, ranging from four to ten, the same at every level: below that range the cost of an extra layer outweighs the benefit, and above it the manager no longer has time for proper one-on-ones. Will Larson’s guide to sizing engineering teams gives similar numbers: a manager should support six to eight engineers, and a manager of managers four to six managers.

Work it through. At a span of seven, one founder can directly lead seven managers, who can lead about forty-nine people. Past about sixty, either the founder has more direct reports than anyone can lead, or a second layer appears. Molly Graham, in First Round Review’s account of scaling, describes thirty to fifty people as the point where a company goes from being a family to being a company. The second layer is the structural form of that change.

At the defaults, a hundred and twenty people at a span of seven, the company has about fifteen first-line managers and three managers of managers, and the founder leads only the second layer plus whoever else reports directly. Drag the headcount to fifty and the managers of managers vanish: every manager reports to the founder. Drag the span to ten and the second layer arrives later, at the price of managers too stretched to coach. Drag it to four and layers multiply, each one a step every decision must climb. The figure is a simplification, since founders keep some specialists as direct reports, but the point holds: past about sixty people, the founder either builds this layer deliberately or becomes it.

What a manager of managers does

A first-line manager makes a team of individuals effective. A manager of managers makes managers effective, which is a different job with different work. They hire managers, and remove the ones who are not working. They coach new managers through their first year, which is where most of them struggle. They set direction across several teams and settle the trade-offs between them, which no single team can settle for itself. They decide headcount and priorities within their area. And they carry information both ways: what the founders intend, down; what is actually happening, up.

Workers silhouetted against a pale sky as they climb a tall steel scaffold. — The second layer builds the structure others work from. Its job is to make managers effective, not to do their teams’ work. Photograph: Soner Arkan · Pexels

What they do not do is the work of the teams. Larson notes that someone supporting fewer than four managers is often underused and tempted to meddle in daily operations. The same temptation applies to anyone promoted from the work itself. Graham’s advice applies here exactly: to grow as fast as the company, you have to give away your job every couple of months. The [founder’s identity lesson](/library/from-builder-to-manager-founders-identity-shift) covers the same shift for the founder.

Stop promoting your best engineer by default

The default promotion in a startup is the best individual contributor becoming the manager. It is usually wrong twice: the company loses its best engineer, or salesperson, and gains an untrained manager. At the second layer the cost doubles, because the person now manages managers, a job two steps removed from the skill that got them promoted.

Two changes remove the pressure. First, build a senior technical track, staff and principal levels with pay bands that overlap the management bands, so that promotion does not have to mean management. Larson observes that managers with fewer than four reports tend to function as tech lead managers, part manager and part engineer, and that the role offers limited career growth either way; a real technical track gives such people a better path. Second, choose second-layer managers on evidence of the specific skill: have they made other managers better? Has someone they managed become a good manager? Have they hired a manager who worked out?

Google learned the value of the layer the hard way. In 2002 it removed engineering managers altogether. According to Harvard Business Review’s account, the experiment lasted only a few months, ending when too many people took routine questions, from expense reports to disputes between colleagues, straight to Larry Page. Removing the layer did not remove the work; it moved it to the founder.

The second layer is not a reward for the best individual contributor. It is a different job: making managers good, so the company runs without the founder in every room.

Promote or hire

Promote from within when someone has already done the job informally, coaching a newer manager, running a cross-team project, hiring well, and when the culture of the function matters more than outside experience. Promote with a trial: give the candidate a second team or a new manager to develop for a quarter, with a review date, before the title. A reversible trial avoids the hardest conversation in management, taking back a promotion.

Hire from outside when the function is about to grow past what anyone inside has seen. A head of engineering who has run four teams of eight has done something your best team lead has not. The [first VP lesson](/library/building-leadership-team-first-vp) covers the outside hire. A healthy second layer at a hundred and fifty people usually mixes both, and the people promoted from within should see that the outside hires were chosen for a gap they can name.

Whoever fills the role, write down what it owns before the first day: the teams, the headcount, the budget, the decisions it can make without the founder and the ones it cannot. A manager of managers who has to ask the founder before hiring, before moving someone between teams or before changing a priority is a messenger, and the managers below learn to go around them. Pay the role on the band for its level, with equity that reflects that its decisions now move the outcome; the [compensation bands lesson](/library/compensation-bands-for-startup) has the structure. And give the first ninety days a plan: meet every manager and every manager’s reports in the first month, name the one team that most needs help in the second, and agree with the founder by the end of the third the two or three changes the area will make in the next half. A founder who hands over a layer without these three things has created a title, not a role.

Knowing whether the layer works

Measure managers the way their teams experience them. Google built an upward feedback survey on the behaviours of good managers and, by HBR’s account, saw median favourability rise from 83 to 88 per cent between 2010 and 2012; one vice president’s score rose from 46 to 86 per cent within two survey cycles. A hundred-person startup does not need Google’s survey, but it needs the habit: twice a year, every manager’s reports answer five or six questions about coaching, clarity and support, anonymously, and the results go to the manager and their manager.

Add skip-level conversations: every manager of managers meets each of their managers’ reports once a quarter, without the manager present, to ask what is working and what is not. The aim is not to bypass the first-line manager but to hear what does not travel upward on its own. And watch three numbers: attrition by manager, time to fill roles by manager, and how many first-line managers in the area have been promoted from within. A second-layer manager whose teams keep people, hire fast and grow their own managers is doing the job.

The second-layer review

Every quarter: draw the org chart and count spans; any manager above ten or below four is a decision to make, not a fact to accept. List every first-line manager in their first year and name who is coaching them. Hold the skip-level conversations. Twice a year: run the upward feedback, review it with each manager of managers, and agree one change per manager. Every year, before the hiring plan: project the headcount twelve months out, redraw the layers it implies, and decide now who will fill the second-layer roles, by promotion with a trial or by hire, so the layer is ready before the founder becomes the bottleneck.

Spans and thresholds here are rules of thumb from published practice; the right span for a team depends on how similar and how senior its work is.