Preventive and corrective IT maintenance: definitions, action plan and metrics
Corrective repairs after a failure, preventive avoids it, predictive anticipates it. How to organise SME fleet maintenance, with checks, Windows commands and metrics.
Published 16 November 2026 · 5 min read
In many SMEs, IT maintenance comes down to repairing what breaks. This approach has a hidden cost: every failure leaves an employee unable to work, is repaired in a hurry and sometimes ends with a purchase that was not needed. Preventive maintenance reverses the logic: you check, fix and replace before the failure.
The short answer
Corrective maintenance repairs after the failure. Preventive maintenance avoids it through regular checks (updates, disks, batteries, security). Predictive maintenance goes further: it measures the real condition of devices continuously and triggers the intervention when a metric deteriorates. A well-run fleet combines all three, with as little corrective work as possible. With a tool like LongLife, maintenance becomes predictive without effort: every device is measured continuously, and you are warned before the failure.
Three types of maintenance
| Type | When | Example | Cost to the company |
|---|---|---|---|
| Corrective | After the failure | The disk fails, you replace it and restore the data | High: work stoppage, urgency, data sometimes lost |
| Preventive (schedule) | On a fixed date | Every month, you check updates and disk space | Medium: time spent, even on devices that are fine |
| Predictive (measured condition) | When a metric deteriorates | The battery drops below 70% of its original capacity: you change it before it fails on the road | Low: you intervene at the right time, on the right device |
Corrective maintenance will never disappear: a broken screen or spilt coffee cannot be predicted. The aim is for it to become the exception.
An SME maintenance plan
| Frequency | Action | How |
|---|---|---|
| Continuously | Monitor alerts: disk nearly full, antivirus disabled, repeated crashes | LongLife, continuously |
| Every week | Check that backups succeeded | Backup console |
| Every month | Apply the Patch Tuesday fixes, check devices that are behind | Windows Update, WSUS or Intune |
| Every quarter | Check the condition of disks and batteries, clean up startup programs | LongLife continuously, or the checks below |
| Every year | Dust desktop PCs and fans, review the replacement plan | Workshop or service provider |
| At every employee departure | Restore the device before reassignment | Reset, hardware check |
Preventive checks, one by one
The first two are run in PowerShell as administrator (right-click the Start menu, then "Terminal (Admin)"), the others from Windows Settings.
1. Repair system files
If Windows behaves strangely (crashes, failing updates), start by repairing the system image, then the protected files:
DISM /Online /Cleanup-Image /RestoreHealth
sfc /scannowDISM repairs the Windows component store from Windows Update, then sfc checks each system file and replaces those that are damaged. The message "Windows Resource Protection did not find any integrity violations" means everything is fine.
2. Check the file system without restarting
chkdsk C: /scanThe /scan option analyses the volume online, without blocking the device. If errors are found, Windows will offer a repair at the next restart.
3. Check the condition of the disks
Each disk must be reported as healthy, in the disk properties (Settings > System > Storage > Disks & volumes) or in the manufacturer's tool. A warning means an immediate backup and replacing the disk. To understand wear indicators, see our definition of SMART.
4. Measure battery wear
Windows can produce a battery report that compares the original capacity with the current capacity. The full method is in our guide to testing a laptop battery.
5. See the latest installed updates
In Settings > Windows Update > Update history, look at the date of the most recent one. If it is more than a month old, the device probably has an update problem: follow our guide to Windows Update stuck.
On one device, these five checks take a quarter of an hour. On 80 devices, that is three days of work per quarter. LongLife does them for you, continuously, across the whole fleet.
The metrics to track
| Metric | Definition | What it tells you |
|---|---|---|
| MTBF | Mean time between two failures of the same equipment | The longer it is, the more reliable the fleet |
| MTTR | Mean time to repair, from report to return to service | Measures how responsive maintenance is |
| Incidents per device per month | Number of support requests relative to the fleet | Spots problem models or departments |
| Share of anticipated failures | Failures detected before the user reports them | Measures the shift from corrective to predictive |
| Average fleet age | Average age of devices | To be weighed against their real condition, never on its own |
An example: an SME with 60 devices has 12 support requests a month, 9 of them due to full disks, worn batteries and slow startups. These are three causes that preventive maintenance would have dealt with before anyone complained.
Corrective maintenance done right
When the failure is there, the order matters:
- Back up the data if the device still boots.
- Diagnose: software failure (repairable remotely) or hardware failure (physical intervention)?
- Lend a device to the user if the repair will take more than half a day.
- Repair rather than replace when the faulty part is isolated: battery, disk, memory, keyboard, charging connector. A damaged motherboard can often be repaired by micro-soldering, for a fraction of the price of a new device.
- Record the intervention in the device's history, to spot recurring failures. In LongLife, every intervention goes into the device's health log.
LongLife, turnkey predictive maintenance
LongLife continuously follows the real condition of each device: hardware, operating system, network and security posture.
- Before the failure: a health score per device and a fleet grade from A to F; detectors analyse your fleet continuously and suggest the action, its priority and the cost of doing nothing.
- The right decision: for each device, keep, upgrade or replace, with the cost and CO₂e of each option.
- Remote fixes: signed scripts, launched in a few clicks, for example to clean up startup programs, with a report per device.
- On-site intervention: a work order with a QR code and a technician page, and the history in the health log.
- Hardware repairs: devices can be sent to the OPP! workshop in Nantes for repair, micro-soldering and refurbishment.
The result: fewer failures suffered, planned interventions, and metrics (incidents, anticipated failures) visible in the reports. LongLife works on its own, or alongside a ticketing tool such as GLPI. The free plan covers up to 15 devices. To estimate what predictive maintenance would save your fleet, try the simulator, or book a meeting. For the overall picture, read our guide to IT fleet management.
Sources
- Microsoft, repair a Windows image with DISM: learn.microsoft.com
- Microsoft, chkdsk command: learn.microsoft.com
- ANSSI (French national cybersecurity agency), IT hygiene guide: cyber.gouv.fr
Frequently asked questions
What is corrective maintenance in IT?
It is the intervention that takes place after a failure or malfunction, to get the equipment working again. It ranges from simple troubleshooting (restarting a service, reinstalling a driver) to hardware repair (replacing a disk, a battery, a screen).
What is the difference between preventive and corrective maintenance?
Preventive maintenance acts before the failure, according to a schedule or a measured condition (updates, cleaning, checking disks and batteries). Corrective maintenance acts after the failure. Good preventive maintenance reduces the number of corrective interventions, which cost more and block the user.
What is predictive maintenance?
It is preventive maintenance based on continuous measurement rather than a fixed schedule. You monitor real wear (battery, disk, temperatures, crashes) and intervene when a metric deteriorates, before the user suffers the failure.
Which metrics should you track for IT maintenance?
The most useful are MTBF (mean time between failures), MTTR (mean time to repair), the number of incidents per device per month, and the share of failures detected before the user reports them.
How much longer can your fleet last?
The simulator estimates in one minute the budget and CO₂ saved by extending your renewal cycle.