Zero Trust Architecture for UK SMEs: The Complete Guide
Zero trust is a security model that eliminates implicit trust — every user, device and connection is verified before access is granted, regardless of network location. This guide explains zero trust principles in practical terms for UK SMEs, the Microsoft and cloud-native tools that implement it and a realistic roadmap for adoption.
Nathan Hill-Haimes
Technical Director
Most SMEs do not need a new product to start. They need the controls already inside Microsoft 365 configured properly and watched. This guide explains the principles in plain terms, the exact tools that implement them, and a staged roadmap you can actually follow. If you want this run for you, our managed cybersecurity team and dedicated zero trust services cover both the build and the ongoing management.
Why has the traditional security perimeter collapsed?
The old castle-and-moat model trusted everything inside the network and distrusted everything outside it. The firewall was the wall. That assumption no longer holds: your data lives in Microsoft 365 and SaaS apps outside the perimeter, and your people connect from home networks, cafés and personal devices.
When everyone worked in one office and all data sat on premises, "inside the network" meant something. Today it does not. Cloud services sit outside the firewall by design. Hybrid workers connect from anywhere. Suppliers and contractors are granted access to your systems. The perimeter has dissolved.
The real failure is what happens after a credential is stolen. When an attacker phishes a password or buys it from a breach dump, the perimeter model hands them the same access the legitimate user had — because they are now "inside". Phishing remains the most common breach type reported by UK businesses in the government's Cyber Security Breaches Survey 2025. Zero trust removes the payoff: being inside the network grants nothing on its own.
What are the three principles of zero trust?
Zero trust rests on three principles defined by the NCSC and Microsoft: verify explicitly, use least-privilege access, and assume breach. Together they replace one-time perimeter trust with a continuous, evidence-based decision made at every access request. The NCSC publishes formal design principles for this model.
1. Verify explicitly. Every authentication and authorisation request is judged on its merits, every time. Identity is confirmed with strong authentication, device health is checked against a compliance baseline, and context matters: unusual location, unusual time, an unrecognised device. Each signal feeds a risk decision that grants, challenges or blocks access.
2. Use least-privilege access. People get exactly what their role requires and nothing more. Just-in-time and just-enough-access mean elevated permissions are granted for a specific task and time window, then withdrawn — instead of standing admin rights that become a liability the moment a credential leaks.
3. Assume breach. Build controls as if an attacker is already inside. Segment networks so a single compromised device cannot reach everything, monitor traffic for anomalies, and write response playbooks for confirmed breaches, not just suspected ones. The NCSC's zero trust architecture design principles set out this thinking in full.
How do you build zero trust on Microsoft 365?
For an SME already paying for Microsoft 365, most of zero trust is reachable without buying new tools. The path runs through three components of Microsoft's identity and device stack — Entra ID Conditional Access, Intune and Defender — which share signals to make and enforce access decisions automatically.
Conditional Access (Microsoft Entra ID)
Conditional Access is the policy engine of zero trust in Microsoft 365. It evaluates each sign-in against your rules and decides to allow, challenge or block. Well-built Conditional Access policies can require:
- MFA for all users, or only when sign-in risk is elevated
- Device compliance — the device must be Intune-enrolled and meet your baseline
- Location controls — block or challenge sign-ins from specified countries
- Session limits — unmanaged devices get read-only browser access, not full sync
Microsoft's own documentation explains the engine in detail in its Conditional Access reference.
Intune device compliance
Microsoft Intune device compliance defines what "healthy" means — minimum OS version, disk encryption, active antivirus, screen lock — and enforces it. Conditional Access then uses that compliance status as an input: a non-compliant laptop is blocked from Microsoft 365 apps until it is fixed.
Microsoft Defender for Endpoint
Microsoft Defender for Endpoint supplies endpoint detection and response and feeds a live device-risk score back into Conditional Access. If a device shows signs of compromise, its risk rises and access to sensitive apps is automatically cut until the threat is cleared. Defender plus Barracuda email and network protection is the stack AMVIA runs, monitored by our in-house 24/7 SOC.
Underpinning all three, MFA across Microsoft 365 is the single highest-impact control most SMEs are still missing. Start there.
Where does network segmentation fit in zero trust?
Zero trust is not only an identity problem. Network segmentation enforces "assume breach" at the infrastructure layer by preventing lateral movement. Instead of a flat network where every device can talk to every other device, you place devices in VLANs matched to their function and trust level, so one compromise cannot reach everything.
- Workstations on their own VLAN, with no direct route to servers
- Servers segmented separately, access governed by firewall rules
- Guest and visitor Wi-Fi fully isolated from internal systems
- IoT devices — printers, cameras, smart kit — on a dedicated VLAN
The goal is blast radius. If a finance laptop is compromised, segmentation stops the attacker pivoting to your file server or domain controller. And no — zero trust does not retire your firewall. The firewall still polices traffic at the perimeter and between segments; zero trust adds identity and device checks on top.
What does a realistic zero trust roadmap look like for an SME?
Zero trust is a journey delivered in stages, not a single project. A practical maturity path moves from quick identity wins to device compliance, then to advanced controls over 12 to 18 months. Sequence it so each phase delivers measurable risk reduction on its own.
| Phase | Timeline | Priority controls |
|---|---|---|
| Foundation | 0–3 months | Enforce MFA via Conditional Access; enrol endpoints in Intune; block legacy authentication in Exchange Online; set up DMARC, DKIM and SPF |
| Intermediate | 3–9 months | Conditional Access requiring device compliance; Privileged Identity Management for admins; VLAN segmentation; deploy and monitor Defender for Endpoint |
| Advanced | 9–18 months | Phishing-resistant FIDO2 MFA for high-value accounts; Data Loss Prevention policies; bring third-party access under the same identity and device rules; continuous access evaluation |
Most SMEs that stall do so because the controls are switched on but never watched. Configuration is half the job; ongoing management is the other half. Our Microsoft 365 security team builds and runs each phase, tuning policies as your environment changes so the posture holds.
Zero trust network access vs traditional VPN
| Traditional VPN | Zero trust network access (ZTNA) | |
|---|---|---|
| Access scope | Whole network once connected | Per-application, least privilege |
| Trust model | Trust after one authentication | Continuous verification per request |
| Signals used | Credentials at login | Identity, device health, context, risk |
| Lateral movement | Possible across the network | Contained to the specific app |
| Best for | Legacy remote access | Modern hybrid and remote work |
How Mature Is Your Zero Trust Architecture?
AMVIA can assess your current zero trust maturity against Microsoft's framework and prioritise the implementations that reduce your actual security risk most effectively.
Frequently Asked Questions
Yes. Zero trust is a model, not a single product. For an SME on Microsoft 365, the core controls — Conditional Access, MFA and Intune device compliance — are available within Microsoft 365 Business Premium (from £16.90 per user per month, ex VAT) and need no specialist security tools. Network segmentation needs a managed switch and sensible router configuration, which is achievable for most businesses.
No. A firewall still controls inbound and outbound traffic at the perimeter and between network segments. Zero trust does not replace it — it adds identity verification and device compliance to every access decision and drops the assumption that passing the firewall equals trust. Firewall and zero trust are complementary layers, not competing choices.
No. Conditional Access is a specific Microsoft Entra ID feature that applies zero trust identity principles to Microsoft 365 and connected apps. It is the primary tool for implementing zero trust in a Microsoft environment, but the full model also spans network segmentation, device management, application access and data protection. Conditional Access is a key component, not the whole picture.
A VPN builds an encrypted tunnel and grants network-level access once you authenticate. Zero trust network access grants application-level access based on continuous verification — users reach specific apps, not the whole network, and each request is judged against identity, device and context. ZTNA delivers least-privilege access and far better visibility, making it the preferred remote-access model.
Zero trust limits insider damage through least-privilege access, just-in-time elevation and continuous monitoring. An employee can only reach what their role needs, elevated rights are time-limited rather than permanent, and anomalous access patterns trigger alerts regardless of who is behind them. These controls do not stop insider threats outright, but they sharply limit what a malicious or compromised insider can reach before detection.
Related Reading
Keeping Remote Workers Secure Post-COVID-19
How zero trust principles improve remote access security compared to traditional VPN-only approaches.
UK Cybersecurity Guide for SMEs | Practical Steps
The practical cybersecurity steps UK SMEs should take — many of which implement zero trust principles.
Managed Detection and Response (MDR) | Business Guide
How MDR provides the detection and response capability that completes a zero trust security architecture.