Blog / Guides
The startup founder's guide to people ops: what to put in place at each stage of growth
How people operations evolves at each stage of startup growth, from 1–15 to 75–150 people. What to build, automate, and hire for so the basics stay invisible and your team stays focused on product and customers.
Most founders get people ops wrong in one of two directions: ignoring it until something breaks, or over-investing early in tools the company doesn’t need yet. Ben Horowitz calls this “management debt”: the organizational equivalent of technical debt.
This guide is built from discovery conversations with founders and People Leads, and from trusted sources including Y Combinator, First Round Review, and a16z. It’s structured around what actually breaks, not what the textbook says should happen.
TL;DR
- People ops covers everything that touches the employee lifecycle: time off, contracts, onboarding, payroll prep, and compliance. Done well, it’s invisible. The team ships, you sell, the basics run themselves.
- The right setup at 15 people is completely different from what you need at 50. Getting the timing wrong in either direction is expensive.
- Around 30–75 employees, people ops needs to become a real system. Before that, documented processes are enough. After that, manual setups start producing errors you won’t always catch in time — whether you solve that with an HRIS, an AI-driven workflow, or something else is secondary to actually doing it.
- European leave law is more complex than most founders expect, and most off-the-shelf tools don’t handle it correctly.
- Waiting until something breaks visibly is always more expensive than building the right foundation one stage earlier.
What is people ops?
People operations covers everything that touches the employee lifecycle: hiring, onboarding, time off, contracts, payroll prep, performance, and compliance. AIHR defines it as “a holistic, employee-centric approach to supporting and managing your workforce.” The distinction that matters for founders: traditional HR asks are we compliant? People ops asks does this help the team move faster? HR keeps the company out of trouble. People ops keeps the team building.
People ops by stage: quick reference
| Stage | What breaks | What to do |
|---|---|---|
| 1–15 people | Nothing yet, but nothing is documented either | Lay the basics. Employee records, signed contracts, probation tracking, a simple way to request time off |
| 15–30 people | Memory overload: probation dates slip, leave questions take time, payroll prep creeps up | Document and systematize. One source of truth for employee data, written processes, no more “ask the founder” |
| 30–75 people | Leave law errors, GDPR exposure, payroll reconciliation becoming a half-day job | Make people ops a real function. A dedicated owner, automated routines, compliance and local employment law handled |
| 75–150 people | Reporting gaps, multi-country complexity, pay equity visibility | From operational to strategic. Reporting for leadership, performance infrastructure, automation that frees the People Lead |
What does people ops look like at 1–15 people?
At this stage, the goal isn’t building an HR function. It’s keeping everyone focused on two things: shipping product and selling it.
The founders who get this right treat people ops as a background process: lightweight, mostly automated, nothing that interrupts the team. Time-off requests don’t become a Slack thread. Contracts go out from a template and come back signed the same day. Monthly payroll prep takes minutes, not an afternoon. This is achievable without an HRIS. It just requires a few deliberate decisions made early, while there’s still time to make them calmly.
Fix the roof while the sun shines.
What to put in place:
Employee records. One spreadsheet: name, start date, role, salary, contract location, probation end date. Share it with whoever handles payroll.
Contracts. Keep them in a shared folder, not in someone’s personal Drive. When an employee leaves and asks what data you hold on them, a personal Drive folder is a GDPR problem with no clean answer.
Probation dates. In a calendar, not memory. This is the first thing that slips when headcount grows.
Leave policy. One paragraph: how many days, how to request, who approves. Not a handbook. Make it findable by anyone.
Beyond the operational setup, this is also when company culture forms. The first 10–15 hires don’t just fill roles. They define how the company works. Normalising feedback early, making 1:1s a consistent habit, and being explicit about values costs nothing at this stage and compounds significantly later. As First Round Review puts it, a company can’t scale faster than its founders’ ability to grow.
Get these right and people ops stays out of everyone’s way. The team ships. You sell. When 30 people arrives, and it tends to arrive faster than expected, you’ll have a clean foundation to build on rather than a backlog to untangle.
What changes at 15–30 people?
The first thing that happens is you stop being able to hold it all in your head.
There are too many probation timelines, too many leave balances, too many payroll nuances to track from memory. Nothing is broken yet, but the failure modes are starting to appear.
The most common ones:
- A probation review happens late, or not at all, because no one had it in a calendar
- Someone asks “how many vacation days do I have left?” and the answer requires opening three tabs and doing manual arithmetic
- Payroll prep takes an hour every month instead of fifteen minutes
- A new hire joins and there’s no documented process for their first two weeks
One dimension the list above doesn’t capture: compliance. Contracts negotiated informally, one by one, create legal exposure that isn’t visible until someone leaves or something goes wrong.
First things I look into is always compliance. Work, contract, the basics: whatever can maybe bankrupt a company or put someone in prison. Because they did something illegal they didn't know about.
What to actually do:
This is the stage to think seriously about systems. Not necessarily a full HRIS, but something that removes the manual layer from people ops entirely, whether that’s a lightweight HRIS, an in-house built tool, or AI agent workflows that handle requests and track data automatically. The goal is that nobody on the founding team or in a COO role is spending meaningful time on HR administration. Good systems at this stage free up hours every week for real company building.
One source of truth. Pick one place for employee data and put everything there. A shared spreadsheet works if it’s actually maintained. The discipline matters more than the tool.
A written employee handbook. One source of truth for everything an employee needs to know: how things work, what’s expected, what they’re entitled to. Not because the law requires it at this stage, but because “ask the founder” stops scaling.
A basic onboarding checklist. What needs to happen in week one so the new person isn’t piecing it together themselves. Short and maintained beats long and stale.
When you regularly have to look something up to answer an HR question that used to be in your head, the 30-person inflection is close. That’s the moment the manual stack stops being manageable and starts producing errors.
Why do things break at 30–75 people, and what do you do about it?
This is the real inflection point. Molly Graham, former COO at Quip, puts it plainly: “30 to 50 people is where you go from being a family to being a company, and everything starts to get really hard.” Almost every founder who has been through it describes the same sequence: something that worked fine at 20 people stops working at 35, and by 45 the manual stack is actively generating errors.
The most common breakdowns:
- Leave calculation. Leave law is country-specific and full of edge cases that spreadsheet formulas don’t handle. Nordic markets are particularly complex: Sweden alone grants parents 480 days of paid parental leave per child. Most international tools get this wrong, and employees will notice before you do.
- GDPR exposure. Employee data spread across personal Drives, uncontrolled spreadsheets, and email inboxes is a concrete liability. “It’s in a Google Sheet somewhere” is not a defensible answer to a subject access request.
- Payroll prep. At 20 people it takes an hour. At 50, half a day. Not because payroll is complex, but because reconciling data across disconnected places takes time that compounds with headcount, every month.
- Performance starts to need a real frame. Below the founders, first-time managers are running teams of their own, and “we’ll figure it out” stops working as a way to set expectations, give feedback, and reward what’s actually working.
What to do:
This is when people ops stops being a side task and becomes a real function. The manual stack — spreadsheets, Slack threads, shared Drive folders — is producing real errors and absorbing senior time.
Plan for a People Lead. Y Combinator’s Director of People Ops recommends hiring a dedicated HR person “around 20–30 employees or after a Series A round.” Hire late and the first months become infrastructure work rather than people work.
Whether you address the operational layer with an HRIS, AI-driven workflows, or a maintained in-house setup, the criteria that matter at this stage are the same:
Local employment legislation. As you grow into multiple locations, you need to operate under each country’s employment rules — time off, leave, attendance, statutory entitlements. The cost of getting it wrong scales with every new market. Make sure whichever system handles people ops for you actually covers the countries you hire in, and ask specifically about the edge cases before you commit.
Payroll export. You don’t need native payroll integration. You need a clean monthly export your accountant or payroll partner can use without manual correction.
Setup speed. An HRIS that takes three months to implement is the wrong tool at this stage. If onboarding requires an implementation consultant, keep looking.
HR tool adoption. The best HRIS your team doesn’t use is worse than a spreadsheet. Tools that work inside the software your team already uses daily have structurally higher adoption.
One pattern worth knowing before you buy: companies that choose their first HRIS primarily on price, before 20 employees, almost always re-evaluate it 12–18 months later. The tool covers the basics. But as the team grows, the gaps that weren’t visible at 15 people become real friction at 35. Factor in the evaluation cost when deciding whether to invest now or wait.
HRIS as soon as possible. That is my genuine stance: the better structure of your data, and the earlier you have it, the better for you.
What does people ops look like at 75–150 people?
At this stage, the problem shifts. You’re no longer struggling to track individual employee data. You have a system for that. The gap is now insight: you can’t easily answer questions about the organization as a whole.
Common examples:
- A board meeting is coming up and it takes several hours to produce a clean headcount-by-team breakdown, because the data exists but the reporting doesn’t
- You can’t answer “what’s our salary distribution across levels?” without exporting raw data and building a one-off spreadsheet
- Multi-country expansion has created parallel payroll setups with no unified view: two or three systems, none of which talks to the others
- Performance management has worked informally until now; at this size it stops surfacing enough signal, and you’re operating blind on attrition risk and development gaps
The EU Pay Transparency Directive (Directive 2023/970), in force across all EU member states from 7 June 2026, is turning pay equity reporting from optional to required. If you’re an EU company with more than 100 employees, salary distribution data is no longer a nice-to-have dashboard. It’s a compliance obligation. The implementation deadline was confirmed in December 2025: it was not extended.
What to do:
Reporting. You need reporting that answers organizational questions without an export-and-rebuild cycle. By this stage, whichever system holds your people data should produce these views automatically. At 100 people, the data exists. It’s just inaccessible without manual work, and that gap gets more expensive with every new hire.
Performance infrastructure. Informal check-ins and ad hoc feedback work at 30 people. At 100, without a consistent process, you lose visibility into how things are actually going across the org, and by the time problems surface they’re usually already expensive to fix.
From operations manager to strategic lead. A People Lead hired at 30 to run operations should not still be running operations at 100. Automating the operational layer — leave approvals, onboarding, payroll prep, compliance exports — frees them to work on what actually compounds: org design, culture, manager development, hiring quality. The right tooling is almost always cheaper than a second headcount.
No way for me to provide proper dashboards for my leaders. I have to pull the reports, I have to analyze it in Claude.
Where to go from here?
Whatever stage you’re at, the principle is the same: build for the problems you have now, one step ahead of where they’ll start to bite, and keep the routine parts out of the team’s way.
We’re building Taito.ai for founders, early operators, and people pioneers setting up the basics before there’s a formal HR hire. Slack-native and AI-native: the repetitive routines (leave, onboarding, probation reminders, payroll prep) run in the background, leaving the human side of people ops to you. Up and running in days, not months.
If your setup is starting to show cracks, or you want to get ahead of them, join the waitlist.
FAQ
When should a startup hire its first People Lead?
Most companies land in the same place: the right moment is when people ops is consuming more than 5–8 hours per week of someone senior’s time, or when headcount is approaching 50. Y Combinator’s Director of People Ops puts the threshold at “around 20–30 employees or after a Series A round.” Before that, structured processes are usually enough. After that, the cost of not having a dedicated person starts compounding in ways that aren’t visible until they’re already expensive.
Do you need to have everything in place before hiring a People Lead?
No, but having the basics in place changes what that hire actually gets to work on. If employee data lives in one place, leave tracking runs itself, and onboarding has a documented shape, your first People Lead spends their first months on the work that compounds: hiring quality, manager development, culture, org design. If those foundations don’t exist, the same hire spends their first quarter building infrastructure. An HRIS is one way to get there; AI workflows or a well-maintained system of record are others. The point isn’t the tool. It’s that the basics are running quietly in the background by the time a People Lead walks in.
What’s the best thing a founder can do for people ops early on?
Build the basics one stage earlier than you think you need to, and automate the routine parts as soon as they appear. Employee records in one place, contracts in a shared folder, probation dates in a calendar, a leave policy you can find. None of this takes long, and none of it requires a dedicated person. Done early, it stays out of the way. Leave requests, approvals, and payroll prep run themselves, and the founding team’s time stays where it belongs: on product and customers. The founders who get this right rarely talk about people ops. It just works in the background while they build the company.
Is it okay to manage HR in Google Sheets?
At 1–20 people: yes. At 20–35: it depends. If your team is in one country and your leave rules are simple, you can extend it. If you have employees in multiple countries, or markets with specific leave law requirements, the spreadsheet will start generating errors before you notice. At 35+: the risk of material errors (leave miscalculations, GDPR exposure) starts outweighing the cost of a proper system.
What should a People Lead expect to find when they join a 40-person company?
Realistically: employee data in several places, no single source of truth, leave tracking that’s partly a spreadsheet and partly someone’s memory, and an onboarding process that varies by who was available that week. This is normal, not a failure. What matters is getting a clear picture of what exists in the first two weeks and deciding what to fix in what order, not rebuilding everything at once.
