In-House IT vs Managed IT Support: A Cost and Capability Comparison for UK Businesses

Somewhere between fifteen and sixty staff, almost every UK business has the same argument. The IT has stopped being something the office manager can absorb between other duties, tickets are landing faster than anyone can answer them, and two options are on the table: hire somebody, or buy managed IT support in the UK from a provider. The salary figure looks like the honest number and the monthly fee looks like the expensive one, and that instinct is almost always wrong — not because managed support is cheap, but because a salary is only about two thirds of what an employee actually costs.

This comparison puts both models on the same basis. It works through the fully-loaded cost of an in-house IT hire — employer National Insurance, pension, holiday, sickness, recruitment, tooling and training — against the real cost of a managed service including the things providers charge extra for. It covers coverage hours and what happens at 7pm, the single-point-of-failure problem that sits underneath every one-person IT department, the skill-breadth gap that no single hire can close, and the hybrid model where an internal IT lead is backed by a provider. It ends with a decision framework based on headcount, complexity and growth plan, because the honest answer is that the right model changes as a business changes, and the question is not which is better but which is better for you at your current size.

What each model actually means

An in-house IT hire means an employee on your payroll whose job is your IT. At the sizes where this decision arises, it is normally one person, occasionally two. That person answers the tickets, manages the suppliers, handles the laptops, administers Microsoft 365, and gradually becomes the only individual in the organisation who knows how anything is configured. The role is usually advertised as IT Support Technician or IT Manager depending on the salary band, and the honest description of it is generalist: one person covering endpoint support, networking, identity, backup, telephony, security and supplier management, at whatever depth their background allows.

Managed IT support means a contracted provider delivering an agreed scope for a recurring fee, normally priced per user or per device per month. The provider supplies a service desk covering agreed hours, remote monitoring and management tooling on your endpoints and servers, patching, antivirus and endpoint detection, backup monitoring, an escalation path through first, second and third line, and some quantity of on-site attendance. What is included varies enormously between providers, which is why a per-user price is meaningless without the scope document that sits behind it.

The third model is the one that most UK businesses over about seventy staff eventually land on and few consider early enough: a hybrid, where one internal person owns the relationship, the business knowledge and the day-to-day floor-walking, and a managed provider supplies the out-of-hours cover, the specialist depth, the tooling and the holiday backfill. It is not a compromise between the two; it is a different allocation of work, and when it is set up properly it is usually the model that survives growth best.

The numbers that frame the decision

Four figures explain why the comparison rarely goes the way the salary line suggests. None of them are controversial, and all of them are visible in any UK business’s own payroll data if somebody goes looking.

The first two figures deserve unpacking together, because they compound. A £42,000 salary becomes something close to £56,000 once employer National Insurance, the pension contribution and the ordinary overheads of employing somebody are counted. That £56,000 buys the availability described in the second card — not fifty-two weeks, but around thirty-eight once statutory holiday, bank holidays, realistic sickness and the training days that keep a technical person employable are removed. The effective cost of an available week of that person’s time is therefore closer to £1,470 than the £808 that dividing the salary by fifty-two suggests.

The third card is the comparison figure, and its range is wide for a reason. At the bottom of that band you are buying a reactive service desk with monitoring and patching. At the top you are buying a service that includes endpoint detection and response, a security operations capability, defined project hours, a named account manager and a technical account or virtual CIO function. Providers quoting £45 and providers quoting £115 are frequently both being honest; they are selling different things.

The fourth card is not a cost figure, and it is the one that most often determines the outcome. A single in-house hire creates a person who is simultaneously your entire IT department, your entire IT documentation, your only escalation path and your only holiday cover. The cost of that concentration does not appear anywhere in the budget until the week it becomes the only thing that matters.

In-house IT versus managed IT support, side by side

The comparison below assumes the most common real scenario: a UK business of roughly forty staff weighing one competent generalist IT hire at around £42,000 against a mid-market managed contract at around £70 per user per month. Both columns are costed on the same fully-loaded basis so the numbers are genuinely comparable.

Two things in that table are worth stating plainly. The first is that at forty users the managed contract is not marginally cheaper — it is roughly a third cheaper on a like-for-like first-year basis, while delivering wider coverage hours and eliminating the absence problem. The second is that the in-house column is not losing on cost alone; it is losing on cost while also being narrower in skill and more fragile in availability. That combination is why the answer at forty users is usually straightforward.

It stops being straightforward in both directions. Below about twelve users the managed fee starts to look expensive relative to the actual volume of work, and a part-time arrangement or a block-hours contract often beats both models. Above about eighty users the in-house column improves substantially, because the fixed cost of a salary is being spread across enough people to compete with a per-user fee that keeps rising, and because at that size there is genuinely enough work to keep somebody occupied. The crossover is not a single number; it is a band, and where you sit inside it depends on complexity more than headcount.

Note also that the two columns are not measuring the same risk. An in-house hire concentrates risk in a person; a managed contract concentrates it in a supplier relationship and a scope document. Neither removes risk, and businesses that choose managed support without reading the scope carefully are simply exchanging a problem they understand for one they have not yet met.

Where the money actually goes — the fully-loaded cost of an in-house hire

The chart below breaks a single £42,000 IT hire into its real components, expressed as a percentage of the total first-year cost of employing that person. Every element is an unavoidable cost of employment in the UK or a practical necessity of the role, and the point of showing them proportionally is that the salary line — the only one most businesses budget for — accounts for barely seven pounds in every ten actually spent.

Take these in order, because each has a definite UK number attached to it rather than a rule of thumb.

Employer National Insurance is charged at 15 per cent on earnings above the secondary threshold of £5,000 a year. On a £42,000 salary that is 15 per cent of £37,000, or £5,550. Employment Allowance can offset up to £10,500 of an employer’s secondary Class 1 liability, but most businesses at this size have already consumed it against the rest of the payroll, so the marginal IT hire pays the full amount. It is worth checking your own position with your accountant rather than assuming either way, because it is a five-thousand-pound swing on a single hire.

Pension under auto-enrolment requires a minimum employer contribution of 3 per cent of qualifying earnings, which is the band between £6,240 and £50,270. On £42,000 that is 3 per cent of £35,760, or £1,073. Many businesses pay above the minimum to be competitive for technical staff, and 5 per cent is common, which takes the figure closer to £1,800.

Recruitment is the line most often left out entirely. A recruitment agency placing an IT support role typically charges between 15 and 22 per cent of first-year salary, so £6,300 to £9,240 on this example. Direct hiring avoids the fee but not the cost: advertising, screening, and forty to sixty hours of management time across shortlisting and interviews. And because IT support roles at this level have historically had high turnover, the fee is not a one-off in the way that it is for a finance or operations hire — a business that changes IT person every two and a half years is paying that line more often than it expects.

Tooling is the line businesses discover after the hire has started. A lone technician needs the same platform stack a provider uses: remote monitoring and management, remote access, a ticketing system, documentation, patch management, endpoint detection, and backup monitoring. Bought at the volume of one small business rather than at a provider’s aggregated volume, this runs between £12 and £30 per endpoint per year for the basics and considerably more if endpoint detection and response is included. For forty endpoints, £1,500 to £3,000 a year is realistic, and every pound of it is included in a managed contract by default.

Training and certification keeps a technical person current and employable. Microsoft role-based certification, CompTIA renewals, vendor training on whatever firewall and backup platform you run: £1,500 to £3,000 a year plus the working days spent doing it. Cutting this line is the most common false economy in the model, because an IT person who stops learning becomes a constraint on what the business can adopt.

Add those to the salary and the honest first-year figure for a £42,000 hire lands between £57,000 and £62,000. The steady-state figure from year two, once recruitment falls away, is around £52,000 to £55,000. Neither figure includes the cost of the work that this person will not be able to do — the firewall migration, the Microsoft 365 tenant restructure, the incident at 9pm on a Sunday — which will be bought in from a third party at day rates of £600 to £1,100 as and when it arises.

Capability scoring — what each model covers well and badly

Cost is the argument people have; capability is the argument they should be having. The grids below score the three models against the work a UK business of thirty to eighty staff actually generates. A high rating means the model handles this well by default; a low rating means it will need to be bought, borrowed or lived without.

The pattern in those three cards is the real content of this article. The in-house model is strong precisely where a remote service desk is weak — presence, context, the ability to walk to a desk and notice that somebody is struggling before they raise a ticket. The managed model is strong precisely where a single person cannot be: at 7pm, in August, and at the depth required to design a site-to-site network or respond to a security incident properly. The hybrid is strong almost everywhere and expensive below about sixty users, which is exactly why it tends to be adopted after a business crosses that line rather than before.

Two rows deserve particular attention. Documentation is scored weak for the in-house model not as a criticism of individuals but as an observation about incentives: a single technician who knows the estate has no operational need to write it down, and the work is therefore always the first to be deferred. Managed providers document because their own service model collapses without it — the person answering your ticket on Thursday is not the person who answered it on Monday. That difference is invisible right up until the day your IT person leaves, and then it is the whole story.

Specialist depth is scored variable rather than weak for the in-house model because it depends entirely on who you hired. A former network engineer will handle your firewall and switching properly and struggle with Microsoft 365 identity; a Microsoft-shop generalist will do the reverse. The problem is not competence, it is that modern small-business IT spans at least six disciplines — endpoint, identity, network, security, backup and telephony — and nobody is genuinely current in all six. Our guides on multi-site network design and backup retention policy are both examples of areas where a generalist will produce something that works but not something that survives an audit or an incident.

Side-by-side cost tables at 15, 40 and 90 users

Headcount changes the answer more than any other variable, because one model has a fixed cost and the other has a variable one. The three tables below run the same comparison at three sizes, using steady-state figures from year two so that one-off recruitment and onboarding costs do not distort the picture. Managed pricing is taken at £70 per user per month, which is a realistic mid-market UK figure for a scope including service desk, monitoring, patching, endpoint protection and backup monitoring.

At fifteen users the comparison is not close, and the reason is arithmetic rather than judgement: a fixed £52,000 divided across fifteen people is £318 each per month, which is four times what a managed contract costs. There is also not enough work. Fifteen users generate perhaps six to ten hours of genuine IT work a week, and a full-time hire will spend the remainder either idle, inventing projects, or drifting into non-IT duties — which is how businesses end up with an IT manager who also runs the office move and the phone system and eventually stops doing IT properly at all.

Forty users is the size at which most UK businesses have this argument, and the managed column still wins by a wide margin — roughly £29,000 a year — while also covering the evening, the holiday fortnight and the security incident. The in-house figure here includes £9,000 of bought-in specialist work, which is not padding: at forty users a business typically has a firewall, a site-to-site link or two, a Microsoft 365 tenant with real governance requirements, and at least one line-of-business application with a supplier who needs managing. One generalist will not deliver all of that unaided, and the day rates for the gaps are the same day rates a managed provider absorbs into its fee.

Ninety users is where the picture genuinely changes, and it changes in a way the totals alone do not show. The managed column is still cheapest, but the gap has narrowed relative to what you are giving up: at ninety staff, the value of having somebody physically present, who knows the operations director is presenting to the board on Thursday and that the warehouse scanners always fail after a Windows update, is substantial and does not appear in any table. The hybrid column at £98,800 buys one capable internal lead plus a co-managed contract at roughly half the full per-user rate, and for most businesses at this size that is the arrangement that holds up best over the following three years.

A caveat on all three tables: per-user pricing is not linear in practice. Most UK providers price in bands and most will negotiate below list at forty users and considerably below at ninety. Equally, a provider quoting £45 per user is not offering the scope modelled here, and a business that compares a £45 quote against these in-house totals is comparing a partial service against a complete one. The only fair comparison is scope-matched, which means writing down what you need covered before you ask anyone for a price.

Note too that these tables cost support, not infrastructure. Microsoft 365 licences, connectivity, telephony, cloud hosting and hardware sit outside both columns and cost the same either way — though a provider with volume agreements will often source them more cheaply, and a good in-house person will often rationalise licence waste that a provider has no incentive to touch. If your cloud spend is a live concern, our guide to Azure cost management covers the part of the bill that neither support model directly controls.

A decision gauge — how close are you to the crossover point?

Headcount is the crudest possible proxy for this decision, and it is the one everybody uses. A better instrument scores the underlying drivers: how much work the estate actually generates, how complex it is, how much of it happens outside office hours, and how much tolerance the business has for a single person holding it all. The gauge below shows where a typical UK business of forty to sixty staff scores when we assess it, on a scale where 100 means an in-house team is clearly justified and 0 means a managed contract is the obvious answer.

A score of thirty-three has a consistent composition, and understanding it is more useful than the headline number. Ticket volume usually scores low: forty to sixty users generate fifteen to twenty-five hours of support work a week, which is well under a full-time role. Estate complexity scores moderately — there is normally a firewall, some form of site-to-site or remote access arrangement, a Microsoft 365 tenant with conditional access, a backup platform and one awkward line-of-business application. Out-of-hours need scores low for most office-based businesses and high for anything with shifts, warehousing, hospitality or clinical operations. Risk tolerance for a single point of failure scores low in almost every business that has already lived through an IT person resigning.

The score rises sharply in four situations. A business running its own on-premises servers or a manufacturing line with industrial control systems adds twenty points of complexity that a remote service desk covers poorly. A business with genuine 24-hour operations adds fifteen. A business with a regulatory obligation that requires demonstrable internal ownership — certain financial services and healthcare arrangements — adds ten. And a business with more than about a hundred users adds twenty simply on volume, because at that point the ticket load genuinely fills one or two roles.

Score yourself honestly on those four and a number above sixty is a real signal that in-house or hybrid is the right structure. Below forty, the managed contract is very likely the correct answer and the argument in your business is probably about control rather than cost. Between forty and sixty is the hybrid band, and that is the largest group.

The first twelve months of a managed IT relationship

One reason businesses hesitate over outsourced IT support is that the transition is unfamiliar in a way that hiring is not. Everybody knows what onboarding an employee looks like. Fewer people know what the first year with a provider looks like, and the honest answer is that it is front-loaded with work, uncomfortable around month three, and materially better than the previous arrangement by month six if the provider is any good.

The equivalent timeline for an in-house hire is shorter to describe and not necessarily shorter in practice. Recruitment takes six to twelve weeks from approval to a signed offer, notice periods add one to three months, and a new technician needs two to three months to learn an undocumented estate. Approval to genuine productivity is commonly six to eight months, during which the business is paying for a vacancy and then for a learning curve. A managed provider is answering tickets in week three. That difference in time-to-cover is rarely costed and frequently decisive when the previous IT person has already resigned.

Coverage and service benchmarks

The figures below are what we see when we assess UK businesses at the point they are making this decision — typically twenty to eighty staff, currently supported either by one internal person or by an ad-hoc arrangement with a local firm. They describe the starting position rather than the destination, and the gap between the first three rows and the last three is the whole argument for structured support of either kind.

Read the top and the bottom of that list together. Ninety-four per cent have endpoint protection and nineteen per cent have tested a restore, which tells you that UK SMEs buy the things that come in a box and neglect the things that require a recurring process. That distinction matters here because it maps exactly onto the difference between the two support models. Buying software is a purchasing decision either model handles. Running a monthly patch review, an offboarding checklist and an annual restore test is an operational discipline, and disciplines survive holidays and resignations only when they are owned by a system rather than a person.

The 38 per cent with any documented SLA is the figure that most often surprises directors. Most businesses with an in-house IT person have never written down what response time they expect, which means there is no standard to fall short of and no basis for the conversation when service slips. A managed contract forces that conversation at signature. Whether the provider then meets the SLA is a separate question, and the answer is in the service review pack — which is why the six-month review in the timeline above is not a formality.

The 27 per cent with out-of-hours cover deserves a caution in the other direction. If your business genuinely stops at 5.30pm and nobody works weekends, out-of-hours cover is a benefit you will pay for and rarely use, and it is reasonable to trade it away for a lower fee. The businesses that must not trade it away are those with shift patterns, e-commerce order flows, warehousing, clinical operations, or a month-end that regularly runs past ten at night.

The single-point-of-failure problem

Every argument above is about cost and capability. This one is about what happens on a specific Tuesday, and it is the factor that most often changes a board’s mind after the spreadsheet has failed to.

Eighteen per cent means that in more than four cases out of five, the resignation of one person removes most of what the organisation knows about its own systems. Not the passwords — those are usually recoverable, if awkwardly. What goes is the reasoning: why the firewall rule exists, which supplier holds the domain, what the workaround is for the application that fails on the first of the month, which users have exceptions and why. That knowledge is rebuilt by the successor over six months of discovery, and the business pays for it twice.

The same concentration shows up in smaller ways continuously. A two-week holiday means two weeks of accumulation. A period of sickness means the same. A resignation means a notice period during which motivation is understandably low, followed by a gap of three to five months before a replacement is productive. None of these are hypothetical risks in the way a fire or a flood is hypothetical; they are certainties with unknown dates. A business that hires one IT person has accepted, by arithmetic, that it will have no functioning IT department for some number of weeks every year.

Managed providers do not eliminate this — they have staff turnover too — but they absorb it structurally. The engineer who leaves takes their knowledge of your estate with them only to the extent that it was never in the documentation system, and providers document because their own delivery model depends on it. The practical test to apply to any provider is simple: ask what happens to your service if the engineer who knows you best resigns tomorrow, and ask to see the documentation platform they would hand to the replacement. A provider who cannot show you that is carrying the same concentration risk you are trying to escape.

There is an intermediate mitigation for businesses committed to in-house. Contract a provider for documentation, escalation and holiday cover only — a co-managed arrangement at perhaps a third of the full per-user rate. It is the cheapest insurance available against the failure mode described here, and it is chronically under-bought because the risk has no invoice attached until the day it does.

Mini case study — a 52-person Leeds manufacturer

A precision engineering business in West Yorkshire, 52 staff across an office and an adjoining production unit, had employed one IT manager for nine years. He was competent, well liked, and the only person who understood the estate: two on-premises servers, a firewall he had configured himself, a Microsoft 365 tenant added five years earlier, a backup appliance, and a production scheduling application from a supplier with two staff. The annual cost on the payroll line was £51,000. He resigned in February with one month’s notice to take a role at a larger employer.

What the business discovered in March was the shape of the problem rather than the size of it. Administrative credentials existed but were undocumented; three of the firewall rules could not be explained by anyone; the backup appliance had been reporting success for fourteen months on a job that excluded the finance share after a folder restructure; and the production application ran on a server whose operating system had been out of support for two years, kept alive because an upgrade required supplier work nobody had budgeted for. None of this was negligence. It was one person, fully occupied by tickets, deferring the work that had no deadline until it had been deferred for years.

The business modelled a replacement hire at £48,000 to £55,000, with a realistic six-month lead time to productivity, and compared it against a co-managed contract: a provider covering the service desk, monitoring, patching, backup and out-of-hours, with a half-day a week on site, at £58 per user per month — £36,200 a year. Remediation was quoted separately at £11,400 covering the server migration, the backup rebuild, a firewall rule audit and an identity tidy-up. They took the managed route, with a clause allowing a move to co-managed if they brought an internal lead back in later.

We thought we were buying a cheaper version of what we had. What we actually bought was the first written description of our own systems that has ever existed. The audit document in week two told us more about our IT than nine years of monthly updates had.

Eighteen months on, the outcomes worth reporting are unglamorous. Ticket volume fell by roughly a third in the first six months, almost entirely by fixing recurring causes rather than by users giving up. Patch compliance moved from an unmeasured baseline to a reported 97 per cent. The out-of-support server was migrated in month four, which was the piece of work that had been on the internal to-do list for two years. The finance share has been in the backup set since week three and a restore from it has been tested twice.

Two things did not go smoothly, and they are worth stating because they are typical. The production scheduling application remained awkward — a provider has no more leverage over a two-person software supplier than an employee does, and the expectation that outsourcing would fix that relationship was misplaced. And the loss of physical presence was genuinely felt on the production floor, where operators who had been used to walking to an office found that raising a ticket for a barcode scanner felt like an obstacle. The half-day on site was increased to a full day in month seven to address it, which added roughly £4,800 a year and was, in the operations director’s assessment, the best-value change they made.

The net position is a support cost of about £41,000 against a previous £51,000 salary line that did not include tooling, training or the cost of the deferred work now paid for. The business is clear that the saving is not the point. The point is that the estate is now described, monitored and covered by more than one person, and that when they grow past eighty staff they intend to hire an internal IT lead and move the contract to co-managed rather than replacing it.

The 12-point evaluation checklist

Whichever direction you lean, the quality of the decision depends on doing this work before you commit. Each item below changes the answer materially, and every one of them is cheaper to establish now than to discover in month four.

1. Measure the actual workload. Count tickets and hours for four weeks, including the invisible work — onboarding, offboarding, supplier chasing, licence administration. Under fifteen hours a week, a full-time hire will be underused. Over thirty, a pure managed contract at standard scope will feel thin.
2. List the disciplines your estate requires. Endpoint, identity, network, security, backup, telephony, line-of-business applications. Mark which need genuine depth rather than competence. One hire can cover perhaps two at depth.
3. Establish your real coverage requirement. What hours does the business operate? What actually happens at 7pm if the ordering system fails — does someone lose money, or does it wait until morning? Answer this honestly before paying for cover you will not use, or skipping cover you need.
4. Cost the in-house option fully. Salary plus employer NI at 15 per cent above the £5,000 threshold, plus pension, plus recruitment amortised over expected tenure, plus tooling, plus training, plus the specialist work that will be bought in regardless.
5. Write the scope before you request prices. A per-user figure without a scope document is not a quote. Specify in-scope devices, response and resolution targets, patching, endpoint protection, backup monitoring, on-site attendance, project hours and what triggers an extra charge.
6. Interrogate what sits outside the contract. The most common exclusions are project work, third-party application support, hardware procurement, out-of-hours attendance and anything described as consultancy. Ask for the day rate for each and a realistic annual estimate.
7. Check the escalation path. Who answers first, who is second line, who is third, and what is the route to a vendor. A provider whose third line is one senior engineer has the concentration problem you are trying to leave behind.
8. Ask for the documentation platform. Insist on seeing what would be handed to a replacement engineer. Also establish, in writing, that the documentation of your estate belongs to you and is exportable at the end of the contract. This clause is routinely absent and painful when it matters.
9. Confirm administrative credential ownership. Your tenant, your firewall, your domain registrar, your backup console. The provider should hold delegated access, not sole ownership. Test this by asking whether you could terminate and retain control of everything without their cooperation.
10. Establish the security baseline included as standard. MFA enforcement, conditional access, endpoint detection and response, patch SLA, and whether Cyber Essentials support is in scope or chargeable. If security is entirely an upsell, the base service is a help desk.
11. Agree the service review cadence and its contents. Quarterly as a minimum, with ticket trends, SLA performance, root-cause analysis and a forward roadmap. Volume statistics alone are not a service review.
12. Read the exit terms. Notice period, minimum term, offboarding assistance and its cost, data and documentation handover. A twelve-month term with a three-month notice is ordinary; a thirty-six-month term with automatic renewal and chargeable offboarding is a reason to negotiate or walk.

Common mistakes when comparing the two models

The errors below are the ones we see repeatedly, and all of them lead to a decision that looks correct on the day and unravels within eighteen months.

• Comparing a salary against a contract fee. The single most common error. A £42,000 salary is a £57,000 to £62,000 first-year cost. Comparing the unloaded figure against a fully-inclusive fee overstates the in-house case by roughly a third before any capability argument is had.
• Treating a £45 quote and a £95 quote as the same product. They almost never are. Price the scope, not the headline. If one provider includes endpoint detection and response, out-of-hours and eight project days and another includes none of those, the cheaper quote will cost more by March.
• Assuming one hire covers every discipline. No individual is currently expert in endpoint management, identity, firewalls, backup architecture, telephony and security operations simultaneously. Hiring one person is choosing which two of those you will do well.
• Ignoring the availability arithmetic. Twenty-eight days statutory holiday including bank holidays, plus realistic sickness, plus training days, leaves around thirty-eight working weeks. Nobody plans for the other fourteen, and they arrive every year.
• Outsourcing without an internal owner. A managed contract with nobody inside the business accountable for it drifts within a year. Somebody — operations director, finance director, office manager — must own the relationship, chair the reviews and hold the provider to the scope. This role is perhaps two hours a month and it is the difference between a service and a subscription.
• Expecting a provider to fix supplier relationships. Your line-of-business application vendor will be exactly as responsive to a provider as to you. Outsourcing support does not confer leverage over third parties, and assuming otherwise causes more disappointment than any other single expectation.
• Deciding once and never revisiting. The right model at twenty users is frequently wrong at eighty. Businesses that revisit at every significant headcount change end up in the hybrid model at roughly the right time; businesses that do not end up defending a decision made when they were half the size.
• Letting the incumbent IT person own the evaluation. Understandable and unfair to everyone. The person whose role is under review cannot be the impartial assessor of the alternatives, and putting them in that position is how businesses end up with a decision nobody trusts.

At a glance — in-house versus managed IT support

If you take one row from that table, take the time-to-cover line. Every other factor can be argued either way depending on circumstances. That one is structural: recruiting and onboarding an IT person takes months during which the business is uncovered, and a provider is answering tickets inside a month. When the decision is being made after a resignation rather than in advance of one, it frequently settles the matter on its own.

How Cloudswitched approaches the decision

We support UK businesses under both models and we co-manage alongside internal IT teams, which means we have no useful reason to push one answer. What we do have is the assessment: four weeks of measured ticket volume, a documented estate audit, a fully-loaded cost model for the in-house option against a scope-matched managed quote, and an honest view on where your business sits relative to the crossover band. For businesses that decide to keep or hire an internal IT lead, we provide the documentation, escalation depth and out-of-hours cover that a single person cannot supply alone.

Frequently Asked Questions

Is managed IT support cheaper than hiring an in-house IT person?

At most UK SME sizes, yes — but only when both are costed on the same basis. A £42,000 IT salary carries employer National Insurance at 15 per cent above the £5,000 threshold, a pension contribution, recruitment costs, tooling licences and training, which brings the genuine first-year figure to somewhere between £57,000 and £62,000. A 40-user managed contract at £70 per user per month is £33,600 plus onboarding. Below about twelve users the managed model is dramatically cheaper because the fixed salary is spread across too few people. Above about eighty users the gap narrows and a hybrid arrangement often becomes the better value.

How much does managed IT support cost per user in the UK?

Typical UK pricing runs from around £45 to £120 per user per month, and the range reflects genuinely different products rather than different margins. At the lower end you are buying a reactive service desk with monitoring and patching. At the upper end the scope usually includes endpoint detection and response, security monitoring, defined project hours, on-site attendance and a technical account or virtual CIO function. Per-device pricing is also common at £25 to £60 per device per month and suits businesses where users have several machines. A price quoted without a scope document cannot be compared to anything.

At what size should a UK business hire its own IT person?

Headcount alone is a poor guide, but as a rough band, a dedicated in-house hire starts to make sense somewhere between eighty and a hundred and twenty users for a business of ordinary complexity. The better test is workload and complexity: measure four weeks of genuine IT work including onboarding, offboarding and supplier management, and count how many technical disciplines your estate requires at depth. Under fifteen hours a week, a full-time hire will be underused. On-premises servers, shift operations, industrial control systems or regulatory obligations all pull the threshold down considerably.

What is co-managed or hybrid IT support?

Co-managed support pairs an internal IT lead with a managed provider. The internal person owns floor presence, business context, user relationships and day-to-day triage; the provider supplies out-of-hours cover, specialist depth, monitoring and management tooling, documentation and holiday backfill. It typically costs between a third and a half of a full per-user managed rate on top of the salary. It is the model most UK businesses over about seventy staff settle on, because it fixes the single-point-of-failure problem without giving up the value of somebody being physically present.

What happens when our IT person goes on holiday?

In a single-hire department, nothing happens — and that is the issue. Work accumulates for the duration and is dealt with on return, and anything urgent is either escalated to an ad-hoc third party at day rates or lived with. Statutory holiday is twenty-eight days including bank holidays, and once realistic sickness and training days are added a full-time employee is available for roughly thirty-eight working weeks a year. A business with one IT person has, by arithmetic, accepted around fourteen weeks a year with no functioning IT department unless it has bought cover.

Do we lose control of our IT if we outsource support?

Only if the contract is written that way, which is why credential ownership and documentation ownership belong in the agreement at signature. Your Microsoft 365 tenant, domain registrar, firewall and backup console should be owned by your business with delegated access granted to the provider, never the reverse. The documentation of your estate should be exportable and contractually yours. Apply one test before signing: could you terminate tomorrow and retain full control of every system without the provider’s cooperation? If the answer is no, negotiate that before anything else.

What is usually excluded from a managed IT support contract?

The most common exclusions are project work, third-party and line-of-business application support, hardware and licence procurement, out-of-hours attendance, on-site visits beyond an agreed allowance, and anything the provider classifies as consultancy. Cabling, office moves and major migrations are almost always separate. None of these exclusions are unreasonable, but they should be listed explicitly with day rates attached and a realistic annual estimate, because the difference between a £45 and a £95 per-user quote is usually found entirely in this list.

How long does it take to switch to a managed provider?

Discovery and documentation take one to two weeks, tooling deployment and baselining a further two, and the service desk cutover usually lands in week three to five. Remediation of what the audit uncovers runs through months two and three, and the service stabilises around month four to six. Compare that to recruitment: six to twelve weeks to a signed offer, one to three months of notice, and two to three months for a new technician to learn an undocumented estate — four to eight months from decision to genuine cover.

Will a managed provider handle our line-of-business application?

They will handle the environment it runs in — the server, the operating system, the database, the backups, the network path and the client deployment — and they will log and chase tickets with the software vendor. What they cannot do is exert leverage over that vendor that you do not already have. If your application supplier is small and slow, outsourcing support will not change that relationship. Establish at contract stage exactly where the provider’s responsibility ends and the vendor’s begins, because this boundary causes more disputes than any other.

Does managed IT support include cyber security?

The baseline usually includes endpoint protection, patch management and MFA enforcement, but genuine security capability — endpoint detection and response, security monitoring, incident response, vulnerability management and testing — is frequently a separate tier. Ask specifically whether Cyber Essentials certification support is in scope or chargeable, and treat a contract where all security is an upsell as a help desk rather than a managed service. Independent testing sits outside the contract either way; our guide on what happens during a penetration test covers what that involves.

How do we measure whether the arrangement is working?

Agree the measures before you start, because they cannot be reconstructed afterwards. The useful ones are first-response and resolution against SLA, first-time-fix rate, ticket volume trend, recurring root causes closed, patch compliance percentage, and the proportion of time spent on planned work rather than reactive tickets. Review them quarterly at minimum with a forward roadmap attached. The same discipline applies to any technology investment — our guide to measuring Microsoft 365 Copilot ROI sets out how to build a baseline that finance will accept.

Can we move back in-house later if we outsource now?

Yes, and a well-run transition to a provider makes it easier rather than harder, because you end up with the documentation, asset register and process definitions that a new internal hire would otherwise spend six months rebuilding. The practical requirements are that credential ownership stayed with you, that documentation is exportable, and that the exit terms include offboarding assistance with a stated cost. Most businesses that move back do not move fully — they hire an internal lead and shift the contract to co-managed, which keeps the depth and the out-of-hours cover.

Related reading

More guidance on budgeting, protecting and planning UK business IT:

• Azure cost management: preventing cloud bill shock
• Backup retention policy: how far back you can actually go
• Multi-site network design: connecting branch offices to the cloud
• What happens during a penetration test: a UK business guide
• Microsoft 365 Copilot ROI: measuring what the licences return