The People Operating System: Running HR When You've Flattened the Org

Flatten the Org and HR Gets More Important

A needs-first organization, the kind I argue for in The GEAR Model, organizes around outcomes and small autonomous teams rather than a tall chart of departments and titles. It is faster and leaner. It also quietly removes a scaffold most companies never notice they are leaning on: the hierarchy itself was doing a lot of the people-work. Your manager was your career path, your performance reviewer, the person who recognized your work and the one you escalated to. Take out the layers and those jobs do not disappear. They become homeless.

That is the real failure mode of flat organizations. It is not too little structure; it is a people system that no one explicitly owns. Hiring drifts, because every pod does it differently. Promotion becomes an invisible pecking order that no one can see or argue with. And strong people leave, not because the work is bad, but because no one is tending their growth. The answer is not to add the layers back. It is to treat the people function as a designed system with a clear owner, the same way every need in the org has one.

HR Is an Enabling Team, Not a Gatekeeper

In a needs-first org, the People function is not a control department that approves headcount and files paperwork. It is an enabling team, in the sense Matthew Skelton and Manuel Pais use in Team Topologies: a team whose whole purpose is to raise the capability of other teams, not to sit in the approval path and slow them down. It owns the operating system for people so that the pods do not each have to invent it:

• the hiring scorecards and interview kits every pod hires with
• the level framework that defines what growth means
• onboarding, so a new owner becomes productive fast
• the performance contract and the feedback rituals
• recognition, and dignified exits

The test of a good People function in a flat org is one question: do the pods move faster because it exists, or slower? Enabling, not gating. If HR has become the team that says "not yet" to everyone else, it has quietly rebuilt the hierarchy you were trying to remove.

Hire the Need

If you organize around needs, you hire around them. A typical job description is a list of tasks and a title, which is exactly how the roles no one can fill get written. Hire for the outcome the seat must produce instead. Geoff Smart and Randy Street's Who method is the cleanest version: define the seat by its mission, three to six measurable outcomes it must produce, and the competencies to get there - a scorecard, not a task list. Name the one capability the seat actually needs rather than demanding a person who is great at two opposite things at once; the full argument for why those hybrid seats are unfillable is in The Unicorn Trap. Then predict with a real work sample and a structured interview over an unstructured culture chat, and slot the offer into the level framework on day one.

The People team owns the scorecard library and the interview kits so every pod hires consistently and quickly, without a central gatekeeper. This section is deliberately short because it is the companion to the hiring section in The GEAR Model; the point here is simply that hiring is one component of a system, not a standalone event.

Onboard: The First Ninety Days

Hiring well is wasted if onboarding is an afterthought, and in a flat org it usually is, precisely because there is no manager whose whole job is to absorb the new person. So make onboarding a system, not a welcome lunch. The hiring scorecard becomes the onboarding plan: the outcomes the seat must produce are the new owner's ninety-day targets, paired with a named guide inside the pod, the context and access they need, and an unambiguous picture of the need they now own.

This is not a soft nicety. Widely cited SHRM figures put a structured onboarding process at roughly 50% higher new-hire productivity and materially better retention across the first three years, and Gallup finds employees who experience strong onboarding far more likely to still be there years later. In a needs-first org, onboarding is the moment you actually hand someone ownership. Do it deliberately, or you spend the next two quarters wondering why a good hire is underwater.

Level and Promote: The Score

Growth in a flat org is a level defined by the scope of the need a person can own, plus a documented record of the outcomes they delivered - a score that travels to a future employer, not a title you keep inflating. Give people two tracks at equal pay, so they can advance by deepening ownership or by managing, and scale the ladder to your size: three or four honest levels at thirty people, not the ten rungs a large technology company runs. The full framework, and why the level is more portable than the title, is in The GEAR Model.

The People function owns this and keeps it transparent, because the fastest way to lose a strong person is to leave them guessing what "next" looks like. An invisible ladder is worse than a modest one.

Performance: The Scorecard Is the Contract

Distributed accountability makes performance management harder, not easier. When no single manager owns a person's output, underperformance can hide for a long time. The scorecard is the answer: because the seat was defined by measurable outcomes, performance becomes a concrete conversation about outcomes owned and hit, not a vague annual rating that arrives too late to help anyone.

Two disciplines matter. First, address underperformance early and directly. A person struggling in a seat that demands the wrong capability is usually a role-design problem, not a character flaw - the argument from The Unicorn Trap - and the humane move is to re-scope or re-seat them before it hardens into a firing. Second, protect talent density. Reed Hastings and Erin Meyer's No Rules Rules is blunt about it: adequate performers in a small team quietly lower the standard for everyone around them, which is why Netflix runs a keeper test, asking whether a manager would fight to keep a given person. You do not have to adopt the hard edge of that culture to take the underlying point seriously in a small, high-ownership pod. The People function owns the cadence that surfaces both of these, kindly and early, so that neither a struggling person nor a fading one is left to drift.

Recognition: What a System Can't Manufacture

This is the part a system cannot fully build for you, and it is worth being honest about that. A level framework, clear ownership, and fair pay remove the big de-motivators: being overlooked, being paid unfairly, having no visible path. But they do not, by themselves, create motivation. That comes from being seen - a manager or a peer who notices the work, names it specifically, and connects it to something that matters. Being the single accountable owner of a need is itself a form of recognition; visible responsibility is a status most people value more than a nicer title.

How you deliver the feedback and recognition that make advancement feel real is its own craft, and it is where flat orgs most often fail quietly. I wrote about the mechanics of it in what NVC taught me about leading teams under pressure, and about helping strong people break through their ceilings rather than plateau in The Performance Paradox. The system is necessary. It is not sufficient.

Offboard With Dignity

Exits are harder in a needs-first org, because when a person owns a need, their departure leaves a hole with no manager standing by to absorb it. Two things protect you. The first is continuity: the moment you create ownership, you create the risk of losing it, so a critical need should never sit with a single undocumented person. The pod and the platform team hold enough shared context that a departure is a transition, not a crisis - this is the Resilience need from The GEAR Model, applied to your people.

The second is dignity. People talk, and how you let someone go is watched closely by everyone who stays. Generous, honest, and reasonably fast exits are not only kind; they protect the trust the whole system runs on. It is not an accident that Netflix pairs its keeper test with real severance. An organization that treats leaving badly teaches everyone still there to keep one eye on the door.

The Small-Company Reality, and What This Doesn't Solve

Everything above can read as though it needs a People department. At ten to a hundred and fifty people you do not have one, and pretending otherwise is exactly how the people system gets neglected. The honest version: at small scale the founder owns the people operating system explicitly - it deserves as much of their attention as any of the four needs in the GEAR model - and buys in fractional or part-time HR expertise to build the scorecards, the level framework, and the onboarding, rather than hiring a full-time generalist too early or leaving it to chance. The system does not have to be big. It has to exist and have an owner.

Sources

• Matthew Skelton & Manuel Pais, Team Topologies - enabling teams raise the capability of other teams. teamtopologies.com
• Geoff Smart & Randy Street, Who: The A Method for Hiring - the scorecard (mission, outcomes, competencies). geoffsmart.com
• Reed Hastings & Erin Meyer, No Rules Rules (Netflix) - talent density and the keeper test, paired with generous severance.
• SHRM and Gallup research on onboarding - structured onboarding is associated with materially higher new-hire productivity and stronger three-year retention (widely cited industry figures). shrm.org
• Zappos and Holacracy - about 18% of staff left within a year of the self-management ultimatum; the company later reintroduced managers. The Washington Post
• Dual-track career ladders (Microsoft, Google) - advance as an owner without becoming a manager. Ken Norton
• Haier's RenDanHeYi model - microenterprises organized around user needs, with hiring and profit rights at the front line. mckinsey.com

Related Reading

• The GEAR Model - the needs-first operating model this people system serves. Start here for the structure; this piece is the people layer underneath it.
• The Unicorn Trap - why some roles are unfillable, and how to design a seat for one coherent capability before you hire for it.
• The KPI Trap - the measurement half of the same problem: the wrong scoreboard pulls planners and executors apart.
• What NVC Taught Me About Leading Teams Under Pressure - the feedback and recognition craft a system cannot manufacture.
• The Performance Paradox - helping strong people break through plateaus instead of stalling.