Short answer. Managed IT for a mid-sized company in Algeria in 2027 holds five things. An inventory: workstations, servers, network, accounts, licences, warranties and renewals, in one place. Monitoring: an agent on each machine watches the disk, the backup, the updates, the internet link and the UPS, and alerts before the breakdown. Maintenance: updates applied in the evening, backups verified every day, a restore tried, accounts reviewed. Support: one number, tracked requests and response times written by priority. Steering: a one-page report each month, a replacement plan, and the documentation and passwords in the company's hands. This guide compares the same month run two ways, shows the report the manager receives, and gives the six questions to ask before signing.
It is for the manager of a company with ten to two hundred workstations whose IT depends on a freelance technician, an employee who “knows about computers” or a relative, the one looking for an IT company or an integrator to take the whole thing over, and the founder living in France who wants their Algerian company to run without them looking after it. The notions quoted are those of public references: managed services are described there as a model in which a provider takes ongoing responsibility for a customer's IT for a fee, as opposed to stepping in only after a breakdown; a service-level commitment sets times for taking charge and for restoring service. The incidents, hours and figures in the example are constructed. Prices, clauses and contract length are matters for negotiation and your adviser; this guide describes what the service must contain.
Six real situations, and what managed IT must do
Each row is a sentence heard in a company, and the answer is a way of organising, not one more technician.
Everything rests on one person: what is sometimes called the bus factor, the number of people whose sudden absence stops everything, equals one; managed IT replaces one person with a team, one number, written response times and documentation the company holds. We pay per breakdown, and they come back: whoever is paid per intervention has no reason to make sure there are none; a monthly fee reverses the interest, since every breakdown avoided is time saved for the provider too. Nobody knows what we have: the inventory says how many workstations, how old, on which system, with which licences and warranties, and what expires within ninety days.
The passwords are with the technician: administration access is stored in a vault the company holds, and the provider works with named accounts that can be closed, as in the guide on roles. Updates land in the middle of the day: they are applied in the evening, after the vendors' monthly releases, many of which fall on the second Tuesday of the month; workstations that no longer receive fixes are identified: a PC still on Windows 10 has received none since 14 October 2025, except through a paid extension programme, limited in time. A customer tells us about the outage: monitoring alerts on a disk filling up, a failed backup, a dropped link, a UPS on battery, and the request is opened before anyone calls.
The same month, two ways: nineteen hours stopped, or two
The figure replays the same month on the same machines, break/fix and then managed.
Break/fix: on the 3rd, the server disk is full and invoicing stops for six hours, the time it takes to reach the technician; on the 8th, the accounting PC is infected by an attachment, four hours; on the 14th, the warehouse Wi-Fi goes down, three hours; on the 23rd, an update installs itself in the middle of the day, two hours; on the 27th, the invoice printer refuses to print, four hours. Nineteen hours stopped, spread over five days, and a discovery on the 19th: the backup has been failing for three weeks without anyone seeing it, which is exactly the situation in the guide on ransomware.
Managed: on the 3rd, the “disk at 85 %” alert arrives in the morning and the clean-up is done that evening; updates are applied on the 6th and the 23rd after closing; on the 8th, the attachment is stopped because the PC is up to date and the account protected; on the 12th, a restore is tried; on the 19th, the backup failure is seen that same morning and fixed. Two incidents remain, the warehouse access point restarted remotely in half an hour on the 14th, and the printer on the 27th, replaced by the one upstairs in an hour and a half: two hours stopped. The difference does not come from better hardware: the machines are the same. It comes from nine actions done the day before. These figures are a constructed example; yours are measured by noting, for a month, each outage and how long it lasted.
«The machines are the same. The difference between nineteen hours stopped and two comes from nine actions done the day before.»
Managed IT for a mid-sized company, in five layers
Maintenance is the accented layer: the evening work nobody sees is what removes the next day's breakdowns.
Inventory. Each workstation, server, network device and account, with its age, system, licences and warranty; it is what makes it possible to say, without searching, how many workstations no longer receive fixes and which ones to replace this year. Monitoring. A small program installed on each machine reports its state and allows remote intervention; tools of this kind combine monitoring, alerts, update deployment, remote control and inventory. Alert thresholds are tuned: a disk at 85 %, a backup that has not run, a PC that has not restarted for thirty days.
Maintenance. Updates, backups verified every day and a restore tried as in the guide on backups, former employees' accounts closed, two-step sign-in switched on, network and email kept with the same care. Support. One number, one address, and for each request a number, a priority and a time; in the example, everything stopped: taken in charge within fifteen minutes and restored or worked around within four hours; a team hindered: one hour and one working day; one PC or one question: four hours and three days. Those times belong to the example; yours are written from what an hour of downtime costs. Steering. The monthly report, the review with the manager, the replacement plan, and a file that makes it possible to change provider without rediscovering everything.
The month's report, on one page
The figure is the page the manager receives on the first Sunday of the month.
What was done: twenty-eight workstations out of thirty up to date, the other two waiting for a planned restart; backups verified on all thirty days, and a server restore tried on the 12th, in three hours forty, which gives the manager the real recovery time; thirty-four requests handled, two urgent, nine disruptive, twenty-three simple, thirty-two of them within the written times, the other two explained. What was avoided: nine alerts out of eleven handled before a user called. Accounts: thirty-one active, all with two-step sign-in, one closed after a departure.
What is coming: four workstations are over five years old; two renewals, the antivirus and the domain name, fall within ninety days. Replacement is reasoned as an average: thirty workstations kept five years means six a year, budgeted for, instead of twelve at once on the day they fail; disks, for their part, fail more often as they age, which is one more reason not to wait. What the manager decides: the report ends with one or two decisions to take, not with a list of jobs; ten minutes of reading, once a month, replace the morning phone calls. The same board can live in the manager's dashboard.
«A provider that does not show what it avoided cannot show what it does: the report fits on one page and is read in ten minutes.»
Taking over a company's IT in 2027, in weeks
Four weeks in the example, the first with nothing changed.
Week 1, inventory and stock-take: every machine and every account recorded, licences and warranties listed, passwords recovered from whoever holds them and put in the vault, a restore trial; nothing is touched, things are looked at. Week 2, bringing up to standard: overdue updates, former employees' accounts closed, two-step sign-in switched on, an out-of-reach backup copy, the list of workstations that no longer receive fixes; this is the accented week, because monitoring placed on a disorderly fleet only rings.
Week 3, monitoring and support: agents installed, alerts tuned to raise only what calls for an action, support number and address announced to the team, response times by priority written. Week 4, routine and first report: evening maintenance, first requests tracked, first report, replacement plan for the year. The takeover is done with the person who looked after it, not against them: they are the one who knows why a given setting exists. Several sites, an old server or business software with no vendor lengthen the bringing up to standard.
Six questions before signing
The second is accented: a response time that is not written down does not exist.
- What the manager sees: one page a month, hours stopped, requests and their times, alerts handled before failure, workstations to replace, renewals within ninety days.
- What the company no longer does: wait for one person to answer, pay for a breakdown that comes back, hunt for a password, discover that a backup had stopped running.
- What is measured: hours stopped per month, requests handled within the agreed times, share of workstations up to date, days of verified backup, average age of the fleet.
What is covered: which workstations, servers, network, sites, software; and what is not, so that it is known before the breakdown. How fast: a time for taking charge and a time for restoring or working around, by priority, the hours and days covered, and what happens outside those hours. What is preventive: the list of actions done without being asked, updates, backup checks, restore trial, account review, and how often.
What we get each month: a one-page report and a meeting to read it; a provider that does not show what it avoided cannot show what it does. Who holds the keys: documentation, saved configurations and administration passwords belong to the company, in a vault it holds; the provider has named accounts. How we part: the notice period, the handover of the file, the closing of access; an exit written down from the start is the best sign nobody will want to leave. The clauses themselves are drafted with your adviser.
- What is managed IT, in practice?
- It is entrusting the day-to-day running of your IT to a provider that takes ongoing responsibility for it for a fee: inventory, monitoring, updates, verified backups, support with written response times, monthly report. The difference from a repair call is simple: the repair technician comes when it breaks, the managed provider is paid for it not to break.
- What is the difference between a maintenance contract and on-demand repair?
- On demand, each intervention is paid for and nobody looks between two breakdowns. With a maintenance or managed-IT contract, the provider monitors, maintains and intervenes within written times; it has an interest in there being few breakdowns. The word “maintenance” covers very different offers: read what is preventive, what is measured and what is handed over each month.
- How much does managed IT cost for a mid-sized company in Algeria?
- The fee depends on the number of workstations and servers, the number of sites, the hours covered, the response times asked for and the starting state of the fleet; an initial bringing up to standard is often counted separately. This guide gives no rate. The right comparison is with what hours of downtime cost today: note every breakdown and its duration for a month, and multiply by what an hour of the stopped team is worth.
- Should we hire an IT person or use a provider?
- Up to a few dozen workstations, a provider brings a team, monitoring tools and a continuity one person alone cannot offer, holidays and illness included. Beyond that, an in-house IT person and a provider complement each other: one knows the company and its work, the other holds the monitoring, the evenings, the rare subjects. In both cases, documentation and passwords belong to the company.
- What should an IT maintenance contract contain?
- Six answers: what is covered and what is not; times for taking charge and restoring by priority, with the hours covered; the list and frequency of preventive actions; the monthly report; ownership of documentation and passwords; the exit conditions. The legal clauses are drafted with your adviser; these six points are what should be found in them.
- Our workstations are still on Windows 10: is that a problem?
- Yes. Support for Windows 10 ended on 14 October 2025; a PC that stayed on it no longer receives security fixes, except through a paid extension programme, limited in time. In 2027 those PCs are a known way in. The inventory says which ones can move to the next system and which are to be replaced, and the replacement plan spreads the spending.
- Will the provider have access to all our data?
- It has the access needed to administer, through named accounts, protected by two-step sign-in and logged; it has no need to read your files. Administration passwords sit in a vault the company holds. Any access can be closed the same day. Confidentiality commitments are written into the contract, with your adviser.
- How do we change provider without losing everything?
- By requiring, from day one, that documentation, configurations and passwords belong to the company. The new provider starts with an inventory and a stock-take, recovers the access, changes it, then takes over monitoring. If the former provider alone holds the keys, the first step is to recover them: that is the only real risk in a change.