Management systems for 20 to 300 people: the minimum that works

A company between twenty and three hundred people needs six management systems, and it needs them in a specific order: ownership and decision rights, management routines, goals, performance, pay rules, and people data. Everything else a larger organisation buys assumes those six already work. The order matters more than the content, because four of the six take their inputs from the two before them.

Somewhere in this range, a company stops being able to run on shared context. Nobody announces it. What happens instead is that a series of things that used to be easy — agreeing priorities, comparing performance, explaining a pay decision — become slow and contested, and each function starts solving the problem locally.

People management by company size

What is missing at fifty people is not what is missing at two hundred. The table below is the short answer for each band: what has to exist, what you can defensibly postpone, and the symptom that tells you the postponement has run out.

Under fifty people the honest answer is that very little formal machinery is needed, and installing it early is a common way to slow a company down. What is needed is that decisions have owners and that the handful of rules everyone relies on are written somewhere findable — which is the argument for a short, real policy set rather than a handbook, covered in HR policies for startups.

The six management systems, in order

The thresholds are indicative and vary with hiring pace, number of locations and how much the leadership team is willing to absorb personally. The dependencies do not vary.

1. Who owns what and who decides what

The cheapest system to build and the one everything else depends on. Every significant outcome gets one accountable owner. Every recurring decision gets a decider, a consulted party and an informed party.

Skipping this and starting with performance is the single most common sequencing error. Performance assessment measures against expectations; if the expectations were never defined, the assessment measures how each manager privately interprets the role.

2. Management routines and expectations

A written statement of what a manager here is accountable for, and the calendar that carries it: one-to-ones, team meetings, goal reviews, escalation paths.

Companies underestimate this because the routines feel obvious. They are not obvious to a first-time manager promoted for individual excellence, and the variance between managers is the largest single source of variance in employee experience at this size.

3. Goal setting

Not a methodology purchase. The requirement is that a person can see how their work connects to the company plan, and that this connection is revisited on a defined cadence rather than annually.

A workable minimum: three to five company priorities per quarter, team goals that clearly serve one of them, and a quarterly review that closes the previous set before opening the next. The review discipline matters more than the format.

4. Performance and level expectations

Once you have roles and goals, performance becomes buildable: expectations by level, an assessment against them with an evidence standard, and one calibration round per function.

The threshold is usually reached when two things are simultaneously true: you have enough people doing comparable work that comparison is meaningful, and you have enough managers that inconsistency between them is visible.

5. Compensation rules

Ranges by level, a stated market positioning, a review cycle, approval thresholds and a recorded process for exceptions.

This is the system companies most often postpone and most often regret postponing, because every month of delay adds another set of individually negotiated salaries that the eventual framework has to accommodate. The cost of building it late is not the design work; it is the correction budget.

6. People data and review cadence

A small number of indicators, reviewed on a defined cadence by the leadership team. Not a dashboard project. Five or six numbers are enough:

• Rating distribution by team and by manager
• Promotion rate and time in level
• Pay dispersion within level
• Regretted attrition, by team and by tenure band
• Time to productivity for new joiners
• People per manager and layer count by function

The value is not the numbers themselves. It is that each one points at a specific component: rating distribution at performance, pay dispersion at compensation rules, regretted attrition at management quality. Data is how you find out which part of the system is failing before someone resigns.

What you can safely skip

Being explicit about this matters as much as the list above. Under three hundred people, most companies do not need a competency library, a nine-box grid, a formal talent review process, an engagement survey platform, or a learning management system.

These are not bad tools. They are tools that assume the six systems above already work, and installing them earlier usually adds administration to an organisation whose actual problem is that nobody wrote down who decides.

Almost every company we assess in this range has built system 4 before system 1, and is paying for it in calibration meetings that cannot reach a conclusion.

How to sequence it in practice

Assume roughly two systems per year if you are building them properly alongside running the business. That is not slow; it reflects the fact that each system needs a full cycle of real use before it is worth refining. A system that has been designed but never used through a complete quarter is still a document.

Start by establishing which of the six you actually have, using the test of whether it exists, whether it is used and whether it is connected to the ones around it. Most leadership teams find they have four in some form and two that exist only as intentions, and the two missing ones explain most of the friction they were describing at the start of the conversation.

That assessment is what a People Diagnostic produces, and it usually reorders the plan the leadership team arrived with. Where the answer is that a system has to be built rather than repaired, that is the work described across our services.