Cybersecurity

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.

NH

Nathan Hill-Haimes

Technical Director

11 min read·Mar 2026

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.

PhaseTimelinePriority controls
Foundation0–3 monthsEnforce MFA via Conditional Access; enrol endpoints in Intune; block legacy authentication in Exchange Online; set up DMARC, DKIM and SPF
Intermediate3–9 monthsConditional Access requiring device compliance; Privileged Identity Management for admins; VLAN segmentation; deploy and monitor Defender for Endpoint
Advanced9–18 monthsPhishing-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 VPNZero trust network access (ZTNA)
Access scopeWhole network once connectedPer-application, least privilege
Trust modelTrust after one authenticationContinuous verification per request
Signals usedCredentials at loginIdentity, device health, context, risk
Lateral movementPossible across the networkContained to the specific app
Best forLegacy remote accessModern 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