The Zero Trust Myth
Zero Trust has become one of the most abused terms in cybersecurity marketing. Every vendor's product is now Zero Trust. Every framework document references it. And yet most organisations claiming to have implemented Zero Trust have simply bolted MFA onto their existing perimeter model and called it done.
Zero Trust is not a product. It is a security design philosophy that says: never grant access based on network location alone. Every access request must be verified, every time, regardless of where it originates.
The origin is simple: the traditional perimeter model assumes that anything inside the corporate network can be trusted. Attackers learned to exploit this 20 years ago — by getting inside via phishing, VPN credential theft, or supply chain compromise, they move freely. Zero Trust assumes the perimeter has already been breached.
The Three Core Principles
Every access request must be authenticated, authorised, and continuously validated — regardless of network location. Being on the corporate Wi-Fi grants zero implicit trust.
Grant the minimum permissions required for the specific task, for the minimum time necessary. Just-in-time access, time-limited tokens, automatic expiration.
Design every control assuming that an adversary is already inside. Limit blast radius through segmentation, monitor for lateral movement, encrypt everything in transit.
You cannot buy Zero Trust. You design and implement it. Vendors who sell 'Zero Trust solutions' are selling components that contribute to a Zero Trust architecture — not the architecture itself.
The Six Pillars — CISA ZT Maturity Model
| Pillar | Traditional | Initial ZT | Advanced ZT | Optimal ZT |
|---|---|---|---|---|
| Identity | Password-only | MFA everywhere | Phishing-resistant MFA | Continuous risk-based auth |
| Device | Domain join | MDM enrolled | Compliance health check | Continuous attestation |
| Network | Flat VLAN | Basic segmentation | Micro-segmentation | Dynamic policy per request |
| Application | VPN access | App-level auth | ZTNA per application | Just-in-time provisioning |
| Data | Perimeter DLP | Classification labels | Rights management | AI-driven classification |
| Visibility | Partial logging | Centralised SIEM | Full UEBA | Real-time risk scoring |
Start with Identity — It's Always Identity
If you have limited time and budget, identity controls deliver the highest return per pound invested. The vast majority of breaches involve compromised credentials. Fix identity first.
- Deploy phishing-resistant MFA (FIDO2/WebAuthn or certificate-based) — SMS MFA is better than nothing but is bypassable
- Implement Conditional Access: block sign-ins from non-compliant devices, high-risk locations, and outside expected countries
- Enable Privileged Identity Management (PIM/PAM) — privileged roles should be activated on-demand, not permanently assigned
- Shared accounts are Zero Trust's enemy — every human and every service must have a unique identity
- Remove all legacy authentication protocols (NTLM where possible, basic auth, Kerberos without pre-auth)
- Instrument identity logs into SIEM — impossible travel, new device, off-hours admin access should all alert
# Require MFA + compliant device for all cloud apps
$policy = @{
displayName = "ZT-CA-Require-MFA-CompliantDevice"
state = "enabled"
conditions = @{
users = @{includeUsers = "All"}
applications = @{includeApplications = "All"}
}
grantControls = @{
operator = "AND"
builtInControls = @("mfa", "compliantDevice")
}
}
New-MgIdentityConditionalAccessPolicy -BodyParameter $policyNetwork Microsegmentation
Traditional networks are flat — once inside, attackers move anywhere. Microsegmentation creates software-defined boundaries between workloads, limiting blast radius dramatically.
Start with east-west segmentation between your most critical asset tiers — Domain Controllers, payment systems, and developer environments should never communicate freely with general user workstations.
Phasing Your Implementation
| Phase | Timeline | Focus | Success Metric |
|---|---|---|---|
| Phase 1 — Identity | Month 1–3 | MFA everywhere, privileged access controls, identity logging | 100% MFA coverage, <5 standing privileged accounts |
| Phase 2 — Device | Month 3–6 | MDM enrolment, compliance policies, device health in CA | ≥95% managed device coverage |
| Phase 3 — Network | Month 6–12 | Micro-segmentation, ZTNA for remote access, VPN decommission | VPN replaced, crown-jewel assets isolated |
| Phase 4 — Application | Month 9–15 | App-level auth, JIT access, CASB for SaaS | All apps via ZTNA, no direct internet-to-app exposure |
| Phase 5 — Data | Month 12–18 | Data classification, DLP, encryption everywhere | 100% data classified, DLP on all sensitive categories |
A full Zero Trust implementation takes 18–36 months in a mid-size organisation. Anyone claiming 90-day Zero Trust transformation is selling you a product, not an architecture.
Common Implementation Pitfalls
- Big-bang approach — trying to implement all pillars simultaneously. Phase it. Start with identity.
- Forgetting service accounts — 40% of identities in most environments are non-human. They need Zero Trust too.
- Segmenting without mapping traffic first — micro-segmentation will break things if you don't know what talks to what
- Executive buy-in is non-negotiable — Zero Trust requires changes to how every employee works. This is a business change programme, not an IT project
- Build a rollback plan for every control before deploying — Zero Trust enforcement will inevitably block legitimate access initially
- Measure and report progress quarterly using the CISA ZT Maturity Model — it keeps the programme visible to leadership