How to price MSP services in 2026: a framework for solo and small MSPs

If you run a solo or sub-five-tech MSP and you’re still pricing the way you did in 2020, you’re almost certainly leaving money on the table. The cost base has shifted (EDR, M365 admin, MFA enforcement, backup verification, AI-assisted phishing on every inbox), and so has what the market will pay. This is a practical framework — not a fluffy “value-based selling” deck — for figuring out what to charge in 2026.

Why most MSPs leave money on the table

Most independent MSPs price one of two ways: by copying what the local competition charges, or by working backward from “what feels fair.” Both anchor on numbers from years ago. The result is that a real, fully managed seat — EDR, patching, M365 admin, backup verification, MFA on everything, and helpdesk — gets sold at break-fix margins.

The plumber down the street raises rates every year. The MSP down the street tells a customer, “We haven’t put prices up in three years.” One of these businesses is going to be around in five years and one isn’t.

The three pricing models

There are really only three pricing models in this market. Pick one as your default — mixing too aggressively confuses customers and makes quoting slow.

Per-user (per seat). You price by the number of users you support, regardless of how many devices each one has. This is the cleanest model for white-collar customers — a controller with a laptop, a desktop, and a phone counts as one user. Pros: predictable revenue, simple to sell, M365-aligned. Cons: hurts when a customer’s user-to-device ratio is unusually high (manufacturing floors, kiosks, shared workstations).

Per-device (per endpoint). You price by the count of managed endpoints: workstations, laptops, servers, sometimes mobiles. This works better in environments with lots of unattended devices: shop floors, healthcare clinics, schools. Pros: scales with the actual surface you’re patching and monitoring. Cons: customers can game it by “consolidating” devices off your contract right before renewal.

All-you-can-eat (flat-fee). A fixed monthly amount covers a defined scope, no matter the device or user count. Pros: dead simple, easy to sell to non-technical buyers. Cons: brutal margin compression if the customer grows or has a bad month — and they always have a bad month right after they sign.

For a solo or small MSP, per-user is the cleanest default in 2026, with per-device for industrial environments and all-you-can-eat reserved for very small, very stable customers.

A 2026 reference framework

These are not exact numbers — your market will shift them — but they’re a sane starting point for the US small-MSP segment in 2026, not 2020:

• Fully-managed seat (per-user, all-in): typically $125–$200 per user per month. This is the headline product. Anything materially below that and you’re either subsidizing it from another line or not actually delivering the full stack.
• Co-managed / lite tier (per-user): typically $60–$100 per user per month. Customer keeps an internal IT person; you provide the tooling, backstop, and afterhours.
• Per-device (workstation): typically $75–$110 per managed workstation per month.
• Per-device (server): typically $200–$400 per managed server per month.
• Break-fix block-hours: typically $150–$225 per hour, sold in prepaid 10- or 20-hour blocks. Never sell un-blocked break-fix at 2020 rates — it’s how MSPs go broke politely.
• On-demand hourly (no block): typically $200–$275 per hour. The premium over block-hours is the friction tax for not committing.

If your headline managed-services number is materially below $125/user/mo and you’re delivering the full stack below, you have a pricing problem, not a sales problem.

What to include in the base

The base price has to include everything the customer expects to “just work” — and in 2026 that list is longer than it was five years ago:

• Helpdesk — unlimited remote, within business hours.
• Patching — OS and third-party, with verified reboots.
• Backup verification — not just “we have backups,” but “we restored a file last month and it worked.”
• EDR / managed detection — SentinelOne, Huntress, ThreatLocker, Defender for Business, whichever is in your stack.
• MFA enforcement — not optional, not “we recommend it.” Enforced.
• M365 (or Google Workspace) admin — licensing, group policy, conditional access, shared-mailbox housekeeping.

If any of those is “extra” in your model, you’re going to lose the comparison against the MSP that bundles them in.

What to charge extra for

Bundling everything is a margin trap. These are the line items that should sit outside the base monthly fee:

• Projects — server migrations, M365 tenant moves, network refreshes, anything with a defined start and end.
• After-hours emergency — clearly priced, clearly scoped. A 2 a.m. ransomware call is not “covered in the monthly.”
• On-site visits — bundle a small number per quarter into the base if you must, but anything beyond that is a billable line.
• Hardware procurement — either pass-through with a clear markup, or refer out. Pretending to make money on hardware in 2026 is a slow way to lose customers to Amazon.

The MSPercury approach

This is what the service catalog + package templates + quote engine in MSPercury are built around: each catalog item knows whether it’s per-user, per-device, per-server, or fixed, so a quote built from a 12-question on-site IT Assessment is just a few clicks. Branded PDFs, immutable accepted quotes, no Excel. The point isn’t to be clever — it’s to stop rebuilding the same Essentials / Standard / Premium tier from scratch every time.

Takeaway

• Pick one default pricing model — per-user is the cleanest for most small MSPs in 2026.
• Anchor your fully-managed seat in the $125–$200/user/mo range. If you’re materially below that with the full 2026 stack, you have a pricing problem.
• Bundle the base, bill projects and after-hours separately, and pass hardware through honestly.