MSP backup pricing is usually set once, by instinct, and left alone for years while wholesale storage costs keep falling and client data volumes keep growing in the opposite direction. That mismatch is how backup ends up bundled in free as a loss-leader, or priced so flatly that a 200GB mailbox archive and a 20GB one cost the client the same and cost you very different margins. Getting the model right — not just the number, but which lever you charge on — is what turns backup from a line item nobody looks at into one of the more defensible margins on the invoice.
MSP backup pricing: the three levers, and what each one actually does
MSPs generally price backup on one of three levers, often blended. Per-user or per-device charges a flat monthly rate per seat regardless of how much that seat actually backs up. It's the easiest model for a client to understand and the easiest for you to invoice, but it under-recovers on your heaviest users and over-recovers on your lightest ones (WholesaleBackup).
Per-GB charges for data actually stored. Microsoft's own Microsoft 365 Backup lists at USD 0.15 per GB per month of protected content as one public reference point, and third-party MSP pricing guides quote a broader USD 0.10–0.50 per GB per month range depending on retention length and redundancy (WholesaleBackup; NinjaOne). It tracks real consumption closely, which is the point, but it makes the invoice harder for a client to predict month to month, and it penalises exactly the accounts whose data is growing fastest — often your best clients.
Per-workload charges per protected object: a virtual machine, a Microsoft 365 tenant, a SQL instance. NAKIVO's own published Enterprise Edition pricing starts from roughly USD 3.50 per workload per month (WholesaleBackup), a useful anchor when you're quoting server or VM backup specifically rather than SaaS mailboxes.
None of the three is universally correct. What you're protecting decides it. SaaS mailboxes and file shares usually suit per-user or per-GB. Server and VM backup usually suits per-workload. Anything with wildly uneven data volumes per seat — design teams, finance, engineering — suits per-GB even though it's the least predictable line on the invoice.
The real cost stack: licence, storage and labour
Before setting a sell price, know that your cost has three parts, not one. Licence is what you pay the backup vendor for the software or platform itself. Storage is the wholesale cost of wherever the data physically lands — object storage providers built for backup workloads publish list prices in the low cents-per-GB-per-month range before markup, well under hyperscaler egress-inclusive pricing (WholesaleBackup). Labour is the part MSPs chronically under-price: monitoring jobs for silent failures, chasing the ones that do fail, running periodic restore tests so "backed up" actually means "recoverable", and handling the after-hours call when a client needs a restore right now.
A backup line with a healthy licence-plus-storage margin and zero allowance for labour isn't actually profitable. It just looks that way until the first real incident, when the hours spent on the restore erase months of margin in one ticket.
Margin benchmarks, used as a floor rather than a target
Published industry figures are a useful sanity check even though every book of business differs. Analysis of managed services pricing puts backup services broadly in the 30–60% gross margin range, rising toward 50–70% for well-packaged managed backup specifically — particularly once it's bundled with disaster recovery or ransomware recovery rather than sold as backup alone (GetFlexPoint; NovaBACKUP). That's meaningfully above the 8–18% some MSPs report when backup is priced as a near-pass-through of the vendor's list price with no service wrapped around it (MSP360). The gap between those numbers is the service, not the software. If your backup line sits at the low end, the fix is rarely a bigger markup on the licence — it's charging properly for the labour in the cost stack above, and for the recovery testing most backup contracts quietly include for free today.
Common mistakes that quietly erode margin
A handful of habits show up repeatedly in backup lines that look profitable on paper and aren't in practice. Grandfathered pricing is the biggest one: a rate set for a client three years ago, before their data doubled, that nobody has revisited because the invoice still clears without complaint. Free retention creep is close behind — a client asks for "just a bit longer" retention once, it's granted as a favour, and it quietly becomes the standing policy for that account without the storage cost ever being re-quoted. Restore testing that isn't actually happening is the most dangerous, not because it costs margin today but because it hides the labour cost that should be priced in — the first real disaster recovery, done for real instead of on paper, is when the true cost of that seat finally shows up, usually as an unbilled multi-hour incident. And bundling backup for free to win a managed-services deal treats it as a discount lever rather than a margin line, which makes it very hard to ever start charging for separately later, since the client's baseline expectation is already set at zero.
None of these are pricing-model problems — a well-chosen per-user or per-GB rate does nothing to fix them. They're operational discipline problems that show up on the P&L as a pricing problem, which is why revisiting the model without also auditing which accounts have drifted off it rarely moves the number much.
Packaging backup with security lifts both
Backup increasingly sells better attached to security than pitched alone. A client asking about ransomware protection is really asking two questions — how do we stop it, and how do we recover if we don't — and a single quote answering both closes faster than two separate conversations pitched months apart. It also opens a natural path into compliance work: frameworks like SMB1001 treat backup as one control among several a client needs help meeting anyway, which is a reason to be in the account beyond the monthly backup invoice (see our SMB1001 tier guide).
A worked example
The numbers below are illustrative only, built to show how the pieces fit together — price your own book from your own supplier agreement, not from this article. Say your all-in cost per protected mailbox seat runs to roughly USD 2 a month once licence and typical-volume storage are combined, and you allow another USD 1 a month per seat for monitoring, alerting and amortised restore-testing time. That's a USD 3 cost floor per seat. Pricing at USD 6–8 per seat per month lands inside the 50–60% gross margin band the benchmarks above describe, while still reading as a fair, explainable number next to the cost of doing nothing and losing the data.
That flat per-seat price doesn't mean every seat costs you the same to serve. A client with 400 heavy-data seats and one with 400 light seats will diverge on the storage line even at an identical seat price. Track actual GB consumed per client periodically, even under a flat-rate model, so an account drifting from profitable to break-even shows up before the renewal conversation forces it into the open.
Revisit the model, not just the number
Backup pricing set once and left for five years is common, and it's a mistake, because the two inputs that matter — wholesale storage cost and client data volume — keep moving in opposite directions the whole time. Revisit the model at least annually: which lever you charge on, what the cost stack actually costs today, and whether labour is priced in or quietly absorbed. Run your own numbers through the margin calculator before your next round of renewals, and if you're weighing which vendors to build the stack around, our MSP partner programme covers backup, security and support pricing together.