1. What zero trust means in practice
Zero trust is a way of designing security that stops treating a request as trustworthy simply because it comes from inside the corporate network. Every access to a system, file or application must be verified based on who is asking, from which device, in which context and with which permissions. The network is no longer the main criterion: what counts is the verified identity and the minimum access needed to do one's job.
The term does not refer to a product to install. A supplier may sell you tools that help, but none of them alone makes you zero trust. The real work is a sum of decisions: who accesses what, with what level of verification, with what logging, and who reviews these choices over time. For this reason it is better to talk about a journey rather than a closed project.
2. Why it pays to proceed in phases
A change made all at once almost always blocks the work. If you require everyone to use a new access method on the same day, administration cannot issue invoices, the field engineer cannot open job sheets and support spends its time unlocking users. People find shortcuts, and the shortcuts become the real risk. Proceeding in phases means changing one scope at a time and observing what happens before moving to the next.
Each phase must have a verifiable objective: knowing with certainty which accounts exist, protecting administrative access, isolating a group of systems. If an objective cannot be checked with concrete proof, it is not yet a phase but an intention. This discipline stops you from declaring finished a job that has only just begun, and it makes the plan understandable even to those who decide the budget.
3. First phase: inventory of identities, systems and data
You cannot protect what you do not know. The first step is to build a list of users, service accounts, devices, applications and storage holding important data. For each item, note who is responsible for it, what data it holds and who needs to be able to use it. It is unglamorous work, and it often reveals active accounts of former staff and applications that nobody remembers installing.
You do not need a perfect inventory to start. You need an honest version, with the gaps declared, that is updated every time something changes. Ask your IT supplier to extract the lists from the corporate directory and the cloud services you use, then have the department managers review the result. They are the ones who can say whether an access is still needed, and they can request the revocation of those that are not.
- Extract the list of user and service accounts from the corporate directory and cloud services.
- Assign each system and each storage location to a named owner with a role.
- Flag sensitive data, such as personal, clinical, accounting or contractual data.
- Have managers review access and disable what is no longer needed.
- Save the inventory in a shared document and set a review date.
4. Strong authentication and role-based access
The second phase concerns who gets in. A password alone is no longer enough: multi-factor authentication, meaning a second proof of identity beyond the password, must cover at least email, remote access, administration panels and cloud services. Where possible, prefer methods that are hard to copy, such as a confirmation through a phone app or a physical key. Check with your supplier which applications really support this mechanism.
Next comes the principle of least privilege: each person gets only the permissions their role requires, and administrative powers remain separate from everyday use. An administrator should have a dedicated account for management tasks, separate from the one used to read email. Review groups on a fixed schedule, because permissions tend to accumulate over time and nobody removes them when a role changes.
5. Segmentation and control of internal traffic
Once it is clear who is accessing, you need to limit where they can go. Segmentation divides the network into zones, for example offices, servers, devices and guests, and sets which flows are permitted between one zone and another. A compromised computer in an office should not be able to reach the document server or the backup archive directly. The dedicated guide on network segmentation explains how to design VLANs and the rules that connect them.
There is also a finer version, microsegmentation, which controls traffic between individual workloads. It is useful in complex environments, but it is not the first step for most medium-sized organisations. Before going down to that level, close the obvious paths: direct access between workstations, exposed administration protocols and shared services without control. The comparison between the two approaches is in the dedicated guide.
6. Devices, suppliers and remote access
Device control answers a simple question: is this machine recognised, up to date and authorised to be on this network? Network access control, often referred to as NAC, can block or isolate unknown devices by using authentication at the network port. It is not essential in every business, but it becomes relevant when there are many workstations, devices and guests. The dedicated guide explains how the 802.1X protocol works and when it is worth using it.
Supplier access is often the weakest point, because it is opened for convenience and then forgotten. The comparison between traditional VPN, identity-based access and jump hosts helps you choose according to what must be reachable and by whom. In the meantime, make a list of all active remote accesses and close those that nobody can explain any more. An access closed by mistake is easily reopened; one forgotten and left open is much harder to notice.
7. Monitoring, firewall rules and maintenance
Zero trust without observation makes little sense, because you need to know what is happening. Collect access, authentication and firewall logs in a single place and establish who reviews them and how often. Detection tools such as IDS and NDR can flag unusual traffic, but alerts only have value if someone reads them and turns them into action. Clarify this point before you buy, not after. A log nobody consults is, in practice, a log that does not exist: its usefulness depends on how often someone checks it.
Firewall rules accumulate. Every exception requested for an urgent reason tends to remain for years, and nobody remembers why it exists. A periodic review, carried out with a checklist, reduces unnecessary rules and documents those that remain. It is a maintenance activity to be scheduled as a recurring deadline, not an extraordinary intervention to make when something goes wrong. The owner should receive the reminder in advance, with the list of rules to examine.
8. The cluster guides: what to read and in what order
This guide is the map. Each topic above has a dedicated treatment, designed to answer a specific question. The table indicates where to start depending on your situation. If you do not yet know what you have in the business, start with the inventory; if you already have an inventory and strong authentication, you can move on to the network sections. In every case, write down the choice made and the date, so that the next review starts from a known point.
Do not read all the guides in sequence. Choose the one that addresses the most urgent problem this week, apply what emerges, then return to the map. If third-party access has been open in your business for a long time, start with that, because it is often the intervention with the fastest effect and the least impact on people. Even if you choose a different path, the first move stays the same: understand which accesses exist today before changing them.
| Topic to explore | Question it answers | When to read it |
|---|---|---|
| Network segmentation | How to divide the network into zones and with which rules | When you need to isolate offices, servers and devices |
| Microsegmentation | When finer control than VLANs is needed | When classic segmentation is no longer enough |
| NAC and 802.1X | How to authenticate devices at the network port | When you have many workstations or unknown devices |
| Supplier access | VPN, identity-based access or jump hosts | When you need to open access to third parties |
| Guest network | How to separate Wi-Fi for visitors | When visitors use the internal network |
| Risky protocols | How to close exposed remote access services | When you find services reachable from the internet |
| Attack surface management | What your business looks like from outside | Before setting priorities |
| Firewall review | How to keep rules tidy | Every time the network changes |
9. How to start
The first step this week requires no purchases. Open a shared document and write down the systems that, if they stopped tomorrow, would block the most work. Next to each one, note who manages it and who has administrative access. With this list you have the basis for the first phase, and you can take it to your IT supplier to complete together. Do not aim for completeness on the first attempt: a partial but up-to-date inventory is already more useful than a perfect one left in a drawer.
Then set a date for the first review of administrative access and enable multi-factor authentication on email and management panels, if it is not already active. These are small interventions that do not require a separate project. From there, choose the second topic in the cluster based on what you have discovered, and repeat the cycle: inventory, verification, correction, check. Each round leaves the network a little clearer than before, and it is this result, more than any single tool, that really counts.
Frequently Asked Questions
Is zero trust a product I can buy?
No. Zero trust is a way of organising identities, access, network and controls. Tools can help you implement it, but a product bought without an inventory, clear access rules and designated owners changes little. Be wary of offers that promise complete conversion with a single purchase, and ask your supplier which parts of the work remain your responsibility.
How long does a zero trust journey take?
It depends on the scope: how many systems you have, how different they are, how many suppliers have access and how many people must be involved in the changes. Typically the first useful result comes with the inventory and strong authentication, while segmentation takes longer because it affects the network. Ask your supplier for a phased estimate, with a verifiable objective for each phase.
Do I have to stop using VPN?
Not necessarily, and rarely straight away. A traditional VPN can remain active for some accesses while you introduce identity-based access for other applications. The switch is best done by user groups and by individual applications, verifying that work continues without interruption. The comparison between solutions is in the guide on supplier access.
Is zero trust required by law?
The reference regulations do not use the term zero trust as an obligation. However, GDPR requires security measures appropriate to the risk, and access control is part of that. The NIS2 directive, transposed in Italy by Legislative Decree 138/2024, requires those within its scope to document their management of cyber risks, which usually includes access control. Check with a consultant which measures apply to you.
Where do we start if we do not have an in-house IT manager?
The first task is to appoint a person with the authority to decide on access, even if the technical side remains with an external provider. Without an internal point of contact every change stays suspended between the supplier and management. Ask your supplier to prepare the inventory and present it to someone in the business who can approve exceptions and revocations.
How do I know whether the plan is working?
Check a concrete objective for each phase. Is there a complete list of accounts and owners? Does every administrative account require a second proof? Do suppliers use accesses that are revoked at the end of each intervention? A verbal answer is not enough: you need a document or a log that shows it. If the evidence is there, the phase is closed; if not, the work is not finished yet.
Does zero trust replace the firewall?
No. The firewall remains useful for controlling traffic between zones and towards the outside. What changes is how it is decided whether an access is allowed: not only on the basis of the network address, but also on the identity of the person and the state of the device. The two tools work side by side, and periodic review of firewall rules remains a task to be scheduled.