A compromised Microsoft 365 account, an unmanaged laptop and an over-permissioned supplier connection can each bypass a traditional network perimeter. That is why organisations asking how to implement zero trust should start with access decisions, not a shopping list. Zero trust changes the question from “is this user on our network?” to “should this identity, device and session be allowed to access this specific resource right now?”
For UK organisations, the appeal is clear. Hybrid working, cloud adoption, SaaS sprawl and ransomware have made implicit trust expensive. Yet zero trust is not a product that can be switched on over a weekend, nor is it simply another name for multi-factor authentication. It is an operating model that combines identity controls, device posture, least-privilege access, data protection and continuous monitoring.
The most effective programmes are risk-led and phased. They protect the systems that would cause the greatest operational, financial or regulatory harm first, while avoiding unnecessary friction for employees, partners and customers.
What zero trust means in practice
Zero trust follows a simple principle: never grant broad or permanent trust merely because a user has passed the network boundary. Each access request should be verified against relevant context, such as identity strength, device health, location, sensitivity of the application, unusual behaviour and the requested privilege.
In practice, this does not mean challenging every employee for credentials every few minutes. A well-designed policy applies proportionate controls. A finance director accessing payroll from a managed corporate device may have a low-friction experience. The same account attempting to download payroll data from an unfamiliar device in another country should face additional verification or be blocked.
The aim is to reduce the blast radius of compromised credentials, infected endpoints and misconfigured applications. It also produces better evidence for governance, risk and compliance teams, because access decisions, privileged activity and exceptions become visible and reviewable.
Start with risk, assets and access paths
Before selecting technology, establish what needs protecting and how it is currently reached. Many organisations discover that their most significant exposure is not the core data centre. It may be a cloud file repository containing sensitive client data, an ageing remote-access service, privileged administrator accounts, or a supplier portal with weak controls.
Create a practical inventory of critical applications, data stores, identities, endpoints and third-party connections. For each priority service, document who needs access, from which devices, through which authentication methods and with what level of privilege. This exercise should include service accounts and application programming interfaces, which are often overlooked because they do not resemble conventional user identities.
Risk ranking matters. Protecting every system to the same standard from day one creates cost, complexity and resistance. Start with systems supporting high-value data, financial processes, operational technology, healthcare records or regulated services. Consider the consequences of unauthorised access, data loss, service interruption and ransomware propagation, not just the likelihood of a compromise.
Build zero trust around identity first
Identity is normally the most useful control point because almost every modern attack eventually attempts to impersonate a user, administrator or machine. Centralise authentication where possible and remove separate credentials held in poorly governed applications.
Strong multi-factor authentication should cover workforce access, administrators, remote access and cloud applications. However, not all MFA methods offer the same phishing resistance. Organisations with elevated fraud or ransomware risk should assess stronger methods such as FIDO2 security keys or passkeys, especially for privileged users and finance teams.
Conditional access policies then add context. They can require a compliant device for sensitive services, deny authentication from high-risk locations, block legacy authentication protocols and prompt for step-up verification when activity is unusual. Begin in report-only mode where available, review the impact, and then enforce policies in controlled stages. A badly tested policy can lock out a clinical team, field engineers or a critical supplier as effectively as an attacker can.
Privileged access needs separate attention. Administrators should use dedicated accounts, receive only the permissions required for a defined task and have that access removed afterwards where feasible. Just-in-time access and privileged access management reduce the danger of standing administrator rights. They also make investigations far easier when an incident occurs.
Verify device health and contain endpoint risk
Identity assurance alone is insufficient if the device is infected, unsupported or shared. A zero trust programme should define what qualifies as a trusted endpoint: supported operating system versions, encryption enabled, active endpoint protection, current patches, screen-lock policy and evidence that the device is managed.
Endpoint detection and response can provide valuable device telemetry and help security teams isolate a compromised machine quickly. The right EDR or XDR platform depends on factors beyond detection capability, including the skills available to investigate alerts, existing security tooling, endpoint mix, data residency requirements and whether managed detection and response is required.
Avoid assuming that every personal device must receive the same agent and management profile as a corporate laptop. For some roles, browser isolation, virtual desktops, application-level controls or restricted access to selected SaaS services may be more proportionate. The correct answer depends on the data involved and the business need, rather than a single vendor architecture.
Apply least privilege to applications and data
A large proportion of zero trust work is access hygiene. Remove dormant accounts, excessive group memberships and blanket access to shared drives or SaaS platforms. Then replace broad permissions with role-based access that reflects real job functions.
Role-based access is a useful starting point, but it can become unwieldy in organisations with frequent project work or complex supplier relationships. Attribute-based policies can add useful context, such as department, contract status, device compliance or information classification. The trade-off is greater design and maintenance effort. Policy simplicity is a security control in its own right: if administrators cannot explain why access is granted, they will struggle to manage it safely.
Data classification should inform controls. Sensitive legal files, patient information, financial records and intellectual property may require restrictions on download, sharing, printing or external collaboration. Data loss prevention, SaaS security controls and encryption can help, but accuracy matters. Overly broad rules generate false positives and encourage staff to find workarounds.
Network segmentation remains relevant, particularly for legacy applications, manufacturing environments and critical infrastructure. Segmenting systems limits lateral movement after a compromise. It should complement identity-aware access, not substitute for it. Network location is one signal among many, not proof of trust.
How to implement zero trust in manageable phases
A workable implementation plan usually progresses through five connected phases:
- Assess and prioritise. Map critical services, sensitive data, user groups, devices, suppliers and existing controls. Identify the highest-risk access paths and agree measurable outcomes.
- Strengthen identity. Deploy or improve centralised identity, phishing-resistant MFA where justified, conditional access, privileged access controls and lifecycle processes for joiners, movers and leavers.
- Establish endpoint trust. Define compliance standards, deploy endpoint protection and management, and use device posture as an access signal for important applications.
- Protect applications and data. Reduce excessive permissions, secure SaaS configurations, introduce data-aware controls and segment high-value systems.
- Monitor, test and refine. Feed authentication, endpoint, email and cloud events into appropriate monitoring processes. Run access reviews, test incident response and adjust policies based on real user behaviour.
This sequencing is deliberately flexible. An organisation facing an active ransomware threat may need to prioritise EDR coverage and administrator controls. A professional services firm handling sensitive client documents may gain more immediate value from identity governance, SaaS security and data controls. The programme should reflect the threat model and the operational realities of the business.
Measure outcomes, not product deployment
A zero trust initiative should be governed as a security and business change programme. Metrics should demonstrate reduced exposure rather than simply counting licences deployed. Useful measures include MFA coverage for privileged and standard users, percentage of managed and compliant endpoints, number of dormant accounts removed, proportion of critical applications behind conditional access, standing privileged accounts eliminated, and time taken to revoke access when someone leaves.
Also measure friction. Track failed sign-ins, helpdesk contacts, policy exceptions and the time needed for a new starter or supplier to receive appropriate access. If security policies cause persistent delays, users will seek less controlled routes around them. The objective is controlled access that supports productive work, not security theatre.
Third parties need the same discipline. Supplier accounts should be separately identifiable, limited to the services required, time-bound where possible and reviewed against contractual obligations. Shared credentials should be retired. For high-risk suppliers, request evidence of MFA, endpoint controls, incident notification procedures and access governance as part of the supplier assurance process.
Choose technology with the operating model in mind
The cyber market presents zero trust as a collection of overlapping platforms: identity, EDR/XDR, secure access service edge, email security, data security, SaaS security and security analytics. Buying every category is rarely necessary, and integrating too many point solutions can create its own operational risk.
Assess the capabilities already available in your identity platform, endpoint estate and cloud environment before adding technology. Then identify the gaps that materially affect risk. Procurement should examine integration quality, management overhead, alerting burden, support arrangements, commercial flexibility and how controls work for contractors, legacy systems and branch locations.
An independent comparison can be valuable where several vendors appear capable on a feature list but differ in deployment effort, telemetry quality or day-to-day administration. ITR Cyber helps organisations assess those practical differences against their own requirements, without forcing the decision towards a single manufacturer.
The strongest next move is usually a focused workshop around one high-risk access journey - for example, administrator access to cloud infrastructure or external sharing of sensitive files. Solve that journey properly, measure the result and use the evidence to set the next priority.

