How this managed services pricing calculator builds the price
The price starts from the cost to serve: what one agreement costs to deliver each month. Labour is the modelled hours multiplied by your loaded cost per productive hour, which is the full cost of employing an engineer divided by the hours they actually log to client agreements, not the hours they are contracted for. Add the tool stack, one line per seat for every tool the tier includes plus the tools each server and site needs; an overhead share per seat for the service desk lead, training and internal systems; and a small risk buffer on all of those for the month that goes wrong. That total is the cost to serve.
The floor price is the cost to serve divided by one minus the target agreement margin, rounded up to the next whole pound. Scopable sets out the same shape of formula (labour, tools, shared overhead and a risk buffer, divided by one minus the target margin), and Xero's note on margin and markup explains why the division matters: a markup of half the cost is a margin of only one third. The market check comes after the floor. It can move a price up; it never moves it below the floor.
Per user, per device or a tiered bundle
Price on a unit that moves when your effort moves and that the client can count. For most small and mid-market clients that is the user, because clients think in headcount, and a user's laptop, phone and Microsoft 365 account all come with them. Price servers and sites as add-ons, or a client with a room full of servers gets priced as if it had none. Per device suits estates where devices rather than people drive the work: shared terminals, warehouses, clinics and schools. Whichever unit you choose, never make the service fee a percentage of what the client spends with third parties. That pays you more when the client spends more, whatever it costs you to serve them.
The calculator prices all three models in the same three tiers, so the comparison is about the unit, not the service. Good, better and best should each add something a client can name, such as backups that are tested or someone who plans their IT with them, rather than simply more hours. Steady, Assured and Guided are working names; use your own. The kit's chapter on choosing an MSP pricing model sets out where each unit fits, and the other models beside them.
Reading an agreement once it is signed
Three readings tell you whether an agreement is healthy. The agreement margin is the fee minus the cost to serve, as a share of the fee. The effective hourly rate is the fee, less the tool stack, divided by the hours logged to the agreement: MSPCFO describes it as labour revenue divided by the hours actually delivered, and Pax8's fixed fee effective ratio compares it with your standard rate, treating anything above 1.0 as profitable. The fair-use line is the monthly hours at which the effective hourly rate falls to your rate card. All-inclusive never means unlimited, and a line both sides can see turns a heavy client into a conversation with data, never a surprise invoice.
At DevOpsGroup I would not accept a price on managed work until it cleared a margin floor I had set. The habit I would add from the first day is logging hours against every client, because the reference hours in any model, this one included, are a starting point until your own time records give you better ones.
The annual price review
Labour and tool costs rise every year, and an agreement nobody reprices loses margin quietly until a large catch-up rise puts the client at risk. The review above recomputes every price from next year's costs, which is how the review should be done: the rise is what the model needs to restore the target margin, and never more than the cap written into the contract. MSP Success's January 2025 survey found that 42% of MSPs review prices annually and fewer than one in ten have automatic increases written into the contract, and Compare the Cloud reports UK providers often applying annual increases in the mid to high single digits. Vendor price changes, such as the Microsoft 365 rises from 1 July 2026 reported by The Register, pass through under their own clause, separate from your own review.
A worked example, invented
The defaults describe an invented UK MSP and one invented client, and every figure is illustrative: none is from a real firm, and none is a typical industry value. An engineer costs £35,200 in salary and £5,870 in employer on-costs, £41,070 a year loaded. Of 1,950 contracted hours, holidays, sickness and training leave 1,657.5 available, and utilisation (utilization) of 74% logs 1,226.55 of them to client agreements: a cost per productive hour of £33.48. The client has 46 users, 55 laptops and desktops, 2 servers and 2 sites, and is on the middle tier.
At a 57% target margin and a 5.5% risk buffer, the tiers come out at £69, £94 and £128 per user, with each server at £247 and each site at £88. The client's fee is £4,994 a month for 26.82 modelled hours: an agreement margin of 57.0%, an effective hourly rate of £152.10 against a £120 rate card, and a fair-use line at 33.99 hours. If pay lifts the cost per productive hour by 5.8% and tools rise 7.9%, leaving the price alone takes the margin to 54.4%; recomputed, the fee moves to £5,316, inside an 8.5% cap, and the margin comes back to 57.2%.
Checking the price against the market
Published price bands are wide, and they tell you what others charge, not what you can afford to charge. For the UK, Compare the Cloud's 2026 procurement guide puts a basic tier at £30 to £50 per user a month, a middle tier at £50 to £80 and a top tier at £80 to £150, while Ibertech Solutions says UK SMEs typically pay £50 to £150 per user a month in 2026. Kaseya's 2023 Global MSP Benchmark found $50 to $100 a month the most common per-user band among its respondents. On margin, Service Leadership (ConnectWise) reported an average managed service gross margin of 46.2% for the second quarter of 2024, and Kaseya's metrics guide says there should be at least a 65% gross margin on each agreement. These are all third-party figures, and sources count cost differently, so set your own target on your own definitions. A price well above the bands needs a reason the client will recognise, such as what the tier includes; a price well below them usually means a cost has been left out.
What this does not do
It does not score how hard a client is to support: two clients with the same headcount can need very different effort, and a harder estate needs more hours than the reference. It does not write the service levels, the fair-use rules, the onboarding fee or the contract terms, and it prices one agreement rather than reading every agreement you hold. The managed services pricing guide compares the pricing models in more depth. The Managed Services Pricing and Service Design Kit takes the same method from the price to the promise, with the client effort scorecard, service levels, fair use, onboarding and the annual price review, and it is in development, coming at an introductory £99 + VAT.