How to Start an MSP

What you're actually building

You already know the technology – that's not what decides whether your MSP survives. What decides it is the business machine around the technology: recurring revenue, contracts, pricing discipline, and process. The industry data is sobering: managed-services growth has slowed to roughly 1% worldwide (Service Leadership Index, Q4 2024), and 18% of MSPs ran at a loss that quarter. The same data shows best-in-class shops holding 19%+ adjusted EBITDA – profitability comes from operating discipline, not market tailwinds.

The goal from day one is MRR: fixed monthly fees for defined outcomes, not hours for problems. Recurring revenue is what makes cash flow predictable, hiring possible, and the company eventually sellable – managed recurring revenue is valued at roughly 4–6x annualized, versus 0.5–1x for project work. Every step below points at that goal. Total launch cost typically runs $10,000–$50,000, with a lean solo start near the bottom of that range – see startup costs for the full breakdown, and writing a business plan for putting numbers to your own version.

Step 1: decide the model – side-gig or full-time jump

Side-gig first is the dominant path. Keep the W-2 job and build nights and weekends until the business proves itself. The common jump criterion: side income replacing roughly 75% of your salary, and/or 6–12 months of living expenses banked. Two rules while moonlighting: use a separate laptop (employers can claim IP created on their gear), and honor any non-compete or non-solicit clauses before working your employer's contacts.

The full-time jump trades safety for speed – full-timers typically progress 3–4x faster than side-giggers – but demands the 6–12 months of runway up front, because the revenue curve is slow: a good cold-start first year is $5,000–$10,000 MRR by month 12, and only about 20% of solo founders clear $100k revenue in year one.

Best fit: side-gig if you have a salary to protect and patience; full-time if you have runway, or a warm client base (often an amicable ex-employer) ready to sign.

Step 2: legal entity and insurance

The standard structure is an LLC first, S-corp election later. An S-corp is a tax election, not an entity type; the LLC is cheap with minimal compliance, and you elect S-corp taxation once net profit consistently clears roughly $50,000–$60,000/year (typical savings $2,000–$4,000/year at that level, against $3,000–$8,000/year in added payroll and accounting overhead). Add an EIN, a separate business bank account, and a registered agent.

Insurance is not optional – clients will demand certificates of insurance (COIs) before signing, often naming themselves as additional insured. Typical solo-MSP costs as of 2026: general liability ~$360/year, tech E&O ~$500–$3,000/year, cyber liability $1,200–$5,000/year (priced on revenue, endpoint count, and your own security controls – insurers treat MSPs as high-risk because compromising one MSP cascades to every client). A realistic bundled total: $2,500–$6,000/year. Full detail in legal and insurance.

Step 3: a focused service catalog and a pricing floor

Define exactly what you sell before you sell it: one or two managed plans covering the core – monitoring, patch management, endpoint security, backup, help desk – with everything else explicitly out of scope or quoted as projects. A tight service catalog is what makes fixed-fee pricing survivable.

Price per user at the market norm – typically $100–$250/user/month as of 2026 ($70–$150 at the low end for basic stacks) – and set a floor: a per-client monthly minimum (commonly $500+) and a per-user price you will not go below. The temptation to undercut local competitors by 20% is the classic first-year mistake; it buys you 60-hour weeks at thin margins with no room to hire. Compare structures in pricing models.

Step 4: a lean tool stack

Two pricing camps dominate, and the choice matters for a solo shop:

A realistic month-one stack – RMM/PSA, EDR, backup, email security, password manager, documentation platform (Hudu or ITGlue) – runs $300–$800/month. Watch vendor minimum commitments and 1–3 year contracts: they're a classic cash-flow trap that hits before revenue does. Full comparisons in tool stack and security stack.

Step 5: contracts before the first client

Do not start work on a handshake, and do not download a free MSA template. Spend the $1,500–$5,000 on an attorney-drafted or attorney-reviewed MSA and SOW covering: a defined service description, explicit exclusions (unsupported hardware, EOL software), liability caps, and payment terms – billing monthly in advance. The defined scope is what you point to when refusing out-of-scope work; without it, "all-you-can-eat" means all-they-can-demand. Details and clause-by-clause guidance in contracts and MSA, and set response-time commitments you can actually hit – see SLA design.

Step 6: land the first clients

Warm networks beat every paid channel at the start: referrals convert roughly 3–5x better than cold outreach, and the highest-yield sources are former colleagues, an amicable ex-employer, a hyper-local list of 50–100 small businesses, and partnerships with bigger MSPs handing down their sub-minimum accounts. Lead with a free network assessment as the door-opener, qualify hard, and say no to bad fits – onboarding a client typically costs $10,000–$15,000 with break-even at months 7–12, so one toxic client eats a quarter's profit. The full playbook is in landing your first 10 clients, and picking a vertical early multiplies referrals – see choosing a niche.

Step 7: deliver like a product from day one

The founders who scale treat the service as a product: one standardized stack, one way of onboarding, everything documented. From client one:

• Run a real onboarding process – credentials captured, network mapped, monitoring deployed, baseline hardening applied.
• Document as you go to a standard where a stranger could support the client – see documentation standards. Undocumented environments are unbillable time bombs.
• Standardize ruthlessly. A messy pile of one-off tools and snowflake configurations kills efficiency; define your stack and versions, and migrate new clients onto them.
• Track a handful of numbers monthly – MRR, margin, tickets per endpoint – so problems surface early; see KPIs and benchmarks.

Step 8: the first-year mistakes that kill MSPs

Recurring themes across founder retrospectives – each maps to a step above:

1. Hourly billing. A race to the bottom that attracts clients who undervalue you and rewards firefighting over prevention. Fix: fixed-fee agreements (Step 3), and get past break/fix to ≥50% recurring revenue fast.
2. Underpricing by ~20%. Cheap prices attract cheap clients who fight every ticket, and leave no margin to hire your first technician later.
3. Taking every client. With $10–15k onboarding cost per client, bad fits are quarter-killers (Step 6).
4. No real contracts. No scope, no exclusions, no liability caps – every dispute becomes your loss (Step 5).
5. No standardization or documentation. The efficiency debt that caps you at ten clients (Step 7).
6. No marketing. Referral flow alone stalls; if people don't know you exist, nothing else matters – build a repeatable motion per sales and marketing.

Where to start

This month: form the LLC, open the bank account, get insurance quotes, and engage an attorney on your MSA. Next month: stand up the lean stack, define your one managed plan with its pricing floor, and send the first twenty warm-introduction asks. Within a quarter: sign your first managed client, onboard them by the book, and document everything. Keep the day job until the numbers – 75% salary replacement or 6–12 months runway – say jump. And don't build alone: the MSP world runs on shared knowledge, from peer groups to the forums and podcasts in community resources. The technology was never going to be your hard part – the business machine is, and it's entirely learnable.