Almost everything written about performance management is written for one of two organisations: a thousand-person company with an HR function, or a fifteen-person startup where the founder still knows what everyone did yesterday. The band in between — roughly fifty to four hundred people — is where most Indian MSMEs actually live, and it is where the standard advice fits worst. This is what genuinely changes at around a hundred people, and what to do about it.
The threshold is not headcount. It is visibility.
The number itself is a proxy. The real threshold is the point at which one person can no longer personally observe the work. In a design studio that might arrive at forty. In a single-shift manufacturing unit with a stable supervisory layer, not until two hundred. What matters is whether the promoter is still forming judgements from observation or from reports.
Once judgement comes from reports, two things follow immediately. Whoever writes the reports acquires influence disproportionate to their role. And people who are visible — those who present, escalate, and sit near the founder — start to be rated above people who are merely effective. Neither is anyone’s fault. Both are structural, and both are why a system becomes necessary.
The most common objection at this size is we are only a hundred people, I know everyone. That is usually still true. But “I know everyone” has quietly become “I know the people I see most often”, and the people it disadvantages are typically the ones doing the quiet, essential work that never generates an escalation.
What breaks between fifty and four hundred
| What changes | Why it happens | What it costs if ignored |
|---|---|---|
| A second management layer appears | The promoter can no longer supervise directly | Managers who were promoted for technical skill now manage, untrained |
| Institutional knowledge concentrates | Two or three long-tenured people hold everything | No succession; a resignation becomes an emergency |
| Departments start optimising against each other | Each is measured on something another must sacrifice | Sales promises what production cannot deliver, and both are hitting target |
| Informal pay decisions become visible | Enough people, enough comparison | Resentment that is difficult to answer without a documented basis |
| The founder becomes a bottleneck | Escalation habits formed when the company was 30 | Decisions wait; good managers disengage |
| Hiring outpaces onboarding | Growth | New joiners learn the job by asking whoever is nearest |
What the standard advice gets wrong at this size
Two bodies of advice compete for your attention, and both are addressed to somebody else.
Enterprise advice assumes infrastructure you do not have
Nine-box grids, calibration committees, 360-degree feedback, competency dictionaries. These assume a dedicated HR function, several layers of management, and enough people per role for statistical distribution to mean anything. Applied to a hundred-person business they consume enormous effort and produce ceremony. A forced bell curve across a team of seven is not a distribution; it is an instruction to invent differences. If forced distribution is on the table, the full case against the bell curve at this size is worth reading before you commit to one.
Startup advice assumes a shared context you have outgrown
Continuous feedback, no formal ratings, radical transparency. This works when everyone can see everyone. At a hundred people across two locations, “continuous feedback” without structure means feedback happens where relationships are strongest and not at all elsewhere — which is precisely the inequity a system exists to correct.
What actually fits this band is unglamorous: written result areas per role, a short quarterly conversation, one honest annual review, and evidence attached to ratings. That is the whole of it.
What is genuinely enough at a hundred people
- Five to seven result areas per role, written down. Not per person — per role. Two accounts executives doing the same job should have the same result areas and different targets.
- One measure per result area, with a named source. If the number cannot be produced from something that already exists, either change the measure or build the record first.
- A quarterly conversation of twenty to thirty minutes. What moved, what is blocked, does the target still make sense. Annual-only is too long — a correctable problem becomes a resignation in twelve months.
- Evidence in writing against each rating. This is what makes a promotion defensible a year later, and it is the discipline most often skipped.
- The promoter inside the system, not above it. At this size everyone can see whether it applies upward, and they check.
Notice what is absent: no calibration committee, no competency framework, no 360. Those become useful later, and adopting them now costs attention you need for the basics.
The two mistakes specific to this size
Building for the company you expect to become. A framework designed for four hundred people, implemented at a hundred, is abandoned within a year because the effort is visibly disproportionate. Design for now plus twelve months. Frameworks are cheap to extend and expensive to sustain when oversized.
Underestimating the manager problem. At this stage most managers were promoted for being excellent at the technical work and have never been taught to manage. They will avoid the difficult conversation, rate everyone alike, and describe their teams in adjectives. The framework is rarely what fails — the conversation is. Budget as much attention for manager capability as for design; an LMS is what keeps that capability available to the next person promoted.
The family-business dimension
In many promoter-led Indian MSMEs, some people in scope are related to the promoter, and some long-serving employees hold status disproportionate to their current contribution. Standard advice ignores this entirely; pretending it is not there is what makes systems fail quietly.
Two things help, and neither is comfortable. Have the promoter’s own result areas written and reviewed first, visibly. And decide in advance, in writing, how a family member’s review will be handled and by whom — because the organisation will draw a conclusion from the first such case whether or not you planned one.
Frequently asked questions
At what headcount should we start?
When you notice yourself judging someone’s work from someone else’s account of it. For most organisations that lands between sixty and a hundred and twenty. Starting earlier is not harmful if kept light; starting late means fixing accumulated resentment as well as building the system.
Do we need an HR head first?
No, and waiting for one is a common way to postpone indefinitely. At this size the framework is owned by the promoter and operated by line managers. An HR hire helps with administration, and administration is not the hard part — an HRMS covers most of it.
Should ratings drive increments from year one?
No. In the first cycle managers inflate ratings to protect their teams, and the data is unusable. Decouple for one cycle, say so openly, and link in the second. You will get honest ratings and they will be worth acting on.
How long before it is working?
The first cycle produces a baseline and exposes the design errors. The second is where value appears. Organisations that abandon after year one conclude, wrongly, that performance management does not work — they stopped at the diagnostic.
Where to go from here
If the changes in the table above are recognisable, the framework is the place to start rather than the software. KRA and KPI design covers the definition work, rolling out KRAs covers the twelve weeks after sign-off, and our case study walks through a complete engagement. If you are weighing outside help against buying a system, that decision has its own article.
BPro is based in Kochi, Kerala and works with organisations across India. Call +91 80869 08876, email care@bpropms.com, or use the contact form.