1. Why you need a programme, not a one-off scan
Many businesses run a scan when an alert arrives or when a customer asks for a security statement. The result is a list of problems that nobody assigns, which goes to the IT department and comes back almost identical at the next scan. A one-off scan measures a single moment. A programme measures what happens over time: how many vulnerabilities appear, how many are closed, how long a fix takes and which systems always lag behind.
GDPR requires a level of security appropriate to the risk, and vulnerability management is one of the measures that demonstrates it. The NIS2 directive includes it among the areas to be covered for those within its scope. The ISO/IEC 27001 standard also requires controls over configuration and vulnerabilities. You do not need to cite these regulations in every document, but you do need to be able to show that the process exists, repeats and leaves a trail of the decisions taken.
2. Scope and inventory: know what you have first
You cannot fix what you do not know. The first output of the programme is an inventory of assets: physical and virtual servers, workstations, network equipment, production applications, services exposed to the internet and cloud resources. For each item you need at least a name, an owner, its function, its logical location and its importance to the business. A spreadsheet is enough to start, as long as it is updated every time something comes in or goes out.
The scope must be defined in writing. Inside it are the systems that process personal data, those needed to produce or deliver the service, and those reachable from outside. Outside it may remain decommissioned devices or systems managed by a third party that gives you its own security report. Stating what is out of scope is as important as stating what is in it, because it prevents a system discovered later from being treated as covered by mistake.
3. Roles: who decides, who fixes, who verifies
A programme without owners stalls at the first complex problem. You need at least three distinct roles: whoever runs the scans and collects the data, whoever owns the system and approves the fix, and whoever checks that the process works and that exceptions are justified. In a small business the same person may cover several roles, but the separation should be written down, otherwise nobody notices when something is missing.
If an external supplier runs the scans, its task is precise: deliver the results with a reasoned priority, indicate what requires action from the owner and support fixes on the systems it manages. Ask in writing what it does, how often and in what format it delivers results. A report with no owner and no deadline remains a document, not an action.
4. The six-step cycle
The cycle has six steps, to be repeated on a fixed schedule. Rhythm matters more than the perfection of any single phase. A programme that runs every month, even with simple tools, delivers more value than an ideal one run once a year. When the cycle becomes routine, exceptions surface on their own and responsibilities become clear, because every month someone has to account for what is still open.
- Update the asset inventory and flag what has entered or left the scope.
- Run scans across the scope, preferring authenticated scans where possible.
- Collect the results, separate false positives and record the reason for each exclusion.
- Give each item a priority, an owner and a deadline consistent with the risk.
- Fix it, or apply a temporary mitigation, or record an approved exception with a review date.
- Verify with a targeted scan and close the item only when you have evidence of the fix.
5. Priorities: deciding what to fix first
Scans often find hundreds of findings. Treating them all with the same urgency means not treating any of them well. Priority comes from combining three elements: the technical severity of the flaw, the likelihood that it will be exploited and the value of the affected system. Widely used severity scores exist, along with sources that report flaws already exploited in the wild. They are useful tools, but on their own they are not enough, because a serious problem on an isolated system may matter less than a medium one on an exposed critical service.
A practical rule works in almost every case. First, vulnerabilities actively exploited on exposed and important systems; then serious ones on critical internal systems; then the rest on a predictable schedule. Write the rule on one page, share it with the owners and apply it consistently. Exceptions are approved in writing with a name and a date, not verbally during a meeting.
6. Common mistakes that cause the programme to fail
The most frequent mistakes are not technical. They concern the organisation and discipline of the process, and they recur in almost every business that starts without a written rule. It is worth reading them before you begin and checking them again every quarter, because they tend to return when the team is under pressure and deadlines shorten, which is exactly when the process matters most. A simple check helps: count how many items from the previous review are still open, and why.
- Starting with tools before the inventory, and discovering at the end that the scan does not cover part of the network
- Closing items in the report without a new verification, so the same flaw returns in the next cycle
- Accumulating exceptions with no review date until the register becomes useless
- Fixing only the easy items and leaving those that depend on a supplier untouched
- Forgetting third-party applications, which you often cannot update yourself but can ask your supplier to update
- Treating the programme as a project with an end date rather than as a continuous process
7. How to tell whether the programme works
Metrics should describe how the process behaves, not the absolute number of vulnerabilities found. A rising number can be a good sign, as it may mean the scope has been extended or the scans have improved. What matters more is understanding how many critical items remain open past their deadline, how long a fix takes on important systems and how many exceptions have expired without review.
Present these indicators to the owner every month, with a simple comparison against the previous period. You do not need a sophisticated dashboard. You need a page that the owner can read in a few minutes and on which they can decide: more time, more resources, a change of supplier or the closure of an exception. If the decision never comes, the process has no owner, and it is better to find that out straight away.
8. Topics to explore, and where to start
This guide is an overview of the subject. Each part of the cycle raises practical questions that deserve separate treatment. The first is how to prioritise vulnerabilities using severity scores and sources that report exploited flaws. Next come how to handle out-of-support systems that cannot be updated, how to prepare a penetration test with a clear scope and rules, and how to read a test report and turn it into an action plan.
Other questions concern authenticated versus external scans, the software bill of materials that a supplier may ask you for, the difference between red team, purple team and penetration testing, responsible disclosure of reports received from third parties and the hardening checklist for Windows and Linux servers. If you must choose where to start, start with priority: without a shared rule, every other activity produces work without effect.
9. How to start
This week, do three things. First, list the assets reachable from the internet and the systems that process personal data: they are the minimum scope. Second, assign each one an owner with a first and last name, even if the same person covers several systems. Third, set a monthly review date, with the list of open items in front of you. With these three elements the programme already exists, and the rest is built step by step.
If you entrust scans to a supplier, ask it to describe its process in these six steps. A well-prepared supplier answers concretely about inventory, priorities, verification and reporting. Also ask who approves exceptions and how they are communicated to your owner. If it answers only with the name of a tool, it is not explaining how the problem gets closed.
Frequently Asked Questions
How often should I run scans?
The frequency depends on scope and risk. For systems exposed to the internet it should be higher than for isolated internal ones, and it should be reviewed when applications or architecture change. There is no frequency that suits everyone. Choose one your team can meet every time, and increase it only when your fixing capacity allows.
Is a vulnerability management programme mandatory?
It depends on your sector, size and contractual obligations. GDPR requires security appropriate to the risk, NIS2 may include vulnerability management for those within its scope, and some customers ask for evidence in their questionnaires. To understand what applies to you, ask a legal adviser or your data protection officer, if you have one.
Do I need an expensive tool to get started?
No. An up-to-date inventory, an owner for each system and periodic scanning with a tool suited to your environment are the minimum. More complex tools become useful when the scope grows and the number of items makes manual handling difficult. Before buying, define the process, because the tool must serve the process and not the other way round.
Who must fix vulnerabilities?
The system owner decides and approves the fix, while the IT department or the supplier carries it out. When the fix depends on a third-party product, responsibility also passes through the maintenance contract. If nobody owns a system, that item remains open by definition. That is why the first step is always to assign a name.
What does accepting a risk mean?
It means choosing not to fix a vulnerability for a documented reason, such as a disproportionate cost or a system being retired. The acceptance must be recorded with the name of the approver, the reason, the compensating measures and a review date. A risk accepted without an expiry date is a problem nobody sees. It should be reviewed periodically, not forgotten.
How does the programme link to an incident?
The programme reduces the likelihood of an incident. If one does occur, the vulnerability register helps establish whether the flaw was known and for how long. This matters too when a breach must be notified to the authority, which requires the facts to be reconstructed. Keep the history of scans and fixes, because it is one of the first things requested.
Can I hand everything over to a supplier?
You can outsource running the scans, the analysis and support for fixes. You cannot outsource the decision on risk, which remains with whoever is responsible for the business. Ask the supplier for a written process, an agreed cadence and reports with reasoned priorities. The contract must state who does what and within which response times.