India's flexi-workforce share jumped from 20% of the total workforce in 2020 to 70% in 2025, according to Indian Staffing Federation data. Most founders reading that number see flexibility. Very few of them have re-read their staffing vendor agreement since November 21, 2025 — the day the four new Labour Codes, including the Occupational Safety, Health and Working Conditions (OSH) Code that now governs contract labour, came into force and replaced the framework startups built their hiring habits on.
That gap between "we use contract staffing" and "we know what we signed" is where the expensive mistakes live. Here are the four founders keep making, and what to check before renewing your next SOW.
Your Contractor Count Doesn't Reset Just Because You Split Vendors
The obligation to register as a principal employer has always been calculated per establishment, aggregated across every contractor supplying you workers — not per contractor. If Vendor A staffs your QA team with 8 people and Vendor B staffs your support desk with 9, a labour inspector adds both. Founders who split headcount across two or three staffing firms specifically to "stay small" are not reducing their exposure; they're just making the math harder to check until an inspection forces it.
A Delhi-based logistics-tech company found this out when a routine inspection counted contract workers across housekeeping, security, and IT support contracts they'd never thought to add together. None of the three vendors held documentation the company could produce on demand, because the company had assumed registration was the vendor's problem, not theirs.
Do this: Maintain a single running headcount of every contract worker on your premises or reporting into your systems, regardless of which agency issues their paycheck, and review it quarterly against your registration status.
The Control Test Decides Who's Really a Contractor — Not the Contract
Courts don't read your MSA to decide if someone is genuinely a contractor. They look at who directs the work day to day. In Hussainbhai v. Alath Factory Thezhilali Union, the Supreme Court held that if the principal employer exercises real supervisory control over workers, the arrangement can be treated as employment regardless of what the paperwork calls it.
Here's the fact pattern that trips up startups: a "contract" Salesforce architect who joins your daily stand-up, takes tasking directly from your VP Engineering, uses a company laptop and email address, and has had the same SOW renewed every quarter for over a year. Nothing on paper says "employee." Everything about how the person actually works says otherwise.
Do this: Route day-to-day direction for contract talent through the staffing agency or a documented, deliverable-based scope — not through your own reporting line. Cap SOW renewals and force a real review (convert, extend with new scope, or exit) rather than letting quarterly renewals run on autopilot.
Two Notice-Period Clocks Are Running, and They're Not the Same Clock
Permanent employees in India typically carry a contractual notice period of 30 to 90 days, rising with seniority. Fixed-term or contract engagements are structured very differently — early termination notice on the worker's side is often just one to two weeks. Founders assume this means they can end a contract engagement whenever the project ends, with no runway needed.
What most founders miss is a third clock: the notice-to-vendor clause buried in the staffing MSA itself, which is separate from the worker's own notice terms. Many agency agreements require 30 days' written notice to the agency — or a minimum monthly billing commitment — before an engagement can end, regardless of how quickly the worker themselves could be released.
Before: A startup tells its staffing vendor by phone on a Friday that a data engineer's project wrapped and they won't need the resource from Monday. They expect billing to stop immediately.
After: The MSA specifies 30 days' written notice to the vendor. The startup gets billed for a full month of a resource they've already stopped using, because the notice clock never started on the Friday phone call.
Do this: Read the notice-to-vendor clause before you sign the MSA, not when you're trying to exit an engagement. Match it to how quickly your projects actually end, and always put termination notices in writing on the day you decide, not the day the person's last shift is.
What Contract Staffing Actually Costs vs. Contract-to-Hire vs. Permanent
This is the comparison most founders never run past the first invoice.
| Model | How you pay | Typical India benchmark | Cost pattern over 12 months |
|---|---|---|---|
| Contract staffing | Monthly markup on the worker's pay rate | 15%–35% markup | Compounds every month the role stays open — never resets |
| Contract-to-hire | Monthly markup, then a conversion fee if hired permanently | Markup as above, plus 10%–25% of projected CTC at conversion (often pro-rated against hours already billed) | Higher than either pure model — you pay for the trial period on top of the eventual hire |
| Permanent hiring | One-time recruiter fee, then CTC + statutory employer costs | 8.33%–20% of annual CTC as a one-time fee | Fixed one-time cost, then predictable CTC — no ongoing markup |
Run the actual numbers on a DevOps engineer at a ₹15L CTC (roughly ₹1.25L a month):
Hired permanently through a recruiter, a one-time fee at 15% of CTC costs ₹2.25L. After that, you're paying CTC plus statutory employer contributions — no additional markup stacks on top month after month.
Brought on through a staffing agency at a 25% markup, the same role bills at roughly ₹1.56L a month — about ₹18.75L annualized, which is already more than the CTC itself before you've paid a single recruiter fee, because the markup compounds every single month the role stays open, not once.
The breakeven point for most mid-level tech roles lands around the four-to-five-month mark. Past that, the monthly markup on contract staffing usually costs more over a year than a one-time permanent placement fee, even before you count the eventual conversion fee if you decide to hire the contractor anyway.
Do this: Before choosing a staffing model, estimate how long the role will realistically stay filled — not how long the current project is scoped for. Short, defined engagements favor contract. Anything likely to run past five or six months usually favors direct or contract-to-hire, evaluated on total 12-month cost, not the first invoice.
Which Roles You Legally Shouldn't Keep on Contract for Two Years
Labour law in India has never treated contract labour as a substitute for your core, ongoing workforce. The intent — reinforced across Supreme Court rulings including BHEL Workers Assn. v. Union of India and Gammon India Ltd. v. Union of India — is that contract arrangements used to disguise what is really permanent, central work can be struck down, with courts ordering the workers absorbed as regular employees.
A fintech startup running its entire six-person backend engineering team on the same staffing vendor's payroll for two years, with the SOW renewed quarterly and no defined end date, isn't running a contract engagement. It's running a permanent team through a contracting wrapper — and that's precisely the fact pattern regulators and courts are built to unwind.
Do this: Reserve contract staffing for scope-bound, time-bound, or skill-gap work — a migration project, a seasonal spike, a specialist skill you don't need year-round. If a role is core to your product and has been continuously filled for over a year, move it to permanent or contract-to-hire deliberately, rather than letting renewal cycles decide it for you.
Most of this comes down to reading the MSA before you sign it, not after you're trying to exit one. If you're weighing whether a role should be permanent, contract, or run through an EOR/PEO structure that absorbs the compliance risk entirely, that's a conversation worth having before the requisition goes out, not after the first invoice lands.
