Cybinity
Back to Insights
Zero Trust

NIST ZTA in Practice: What Enterprise Security Teams Get Wrong

Most Zero Trust implementations stall not because of technology, but because of architecture decisions made too early. Here are the most common mistakes enterprise security teams make — and how to avoid them.

By Cybinity Security Team
NIST ZTA in Practice: What Enterprise Security Teams Get Wrong

Mistake 1: Starting with Technology, Not Architecture

The most common mistake is treating Zero Trust as a procurement exercise. A vendor presents a compelling demo, the security team is impressed, and a deployment begins before anyone has defined what the architecture should look like.

The problem is that Zero Trust is an architectural principle, not a product. NIST SP 800-207 defines it as a set of guiding tenets — verify explicitly, use least-privilege access, assume breach — that need to be translated into an architecture before any technology is selected.

Starting with technology means the vendor's product becomes the architecture by default. The gaps in their portfolio become gaps in your security posture.

What to do instead: Begin with a maturity assessment across the CISA Zero Trust Maturity Model's five pillars. Understand your current state, identify your highest-risk gaps, and design the target architecture before engaging vendors.

Mistake 2: Underestimating Identity Governance Complexity

Identity is the right place to start a Zero Trust programme — it's the highest-impact pillar and the most tractable. But security teams consistently underestimate how complex identity governance becomes at enterprise scale.

The challenge isn't deploying an identity provider. It's the work that comes before: understanding who has access to what, why they have it, and whether they still need it. Most enterprises have accumulated years of access entitlements that were never properly reviewed.

This is particularly acute for machine identities — service accounts, API keys, certificates, and workload identities. In most enterprises, machine identities outnumber human identities by a significant margin, and they're typically far less well-governed.

What to do instead: Invest in identity discovery before identity governance. Map your human and machine identities, understand their access entitlements, and identify the highest-risk accounts before implementing controls.

Mistake 3: Treating Microsegmentation as a Network Project

Network microsegmentation is one of the most powerful Zero Trust controls — it limits lateral movement by dividing the network into segments with explicit access controls between them. But security teams often treat it as a network engineering project, handed off to the network team to implement.

Effective microsegmentation requires understanding application dependencies. Before you can segment a network, you need to know which applications communicate with which other applications, over which ports and protocols, and why.

What to do instead: Start with application dependency mapping before implementing microsegmentation. This requires collaboration between security, network, and application teams — and it takes longer than a pure network project, but the outcome is durable.

Mistake 4: Neglecting the Regulatory Mapping

For regulated organisations, Zero Trust isn't just a security improvement — it's a compliance investment. DORA, NIS2, and FCA operational resilience requirements map closely onto Zero Trust controls. But security teams often implement Zero Trust without documenting how their controls satisfy regulatory requirements.

What to do instead: Build regulatory mapping into your Zero Trust programme from the start. For each control you implement, document which DORA articles, NIS2 requirements, or FCA expectations it addresses, and what evidence it generates.

Mistake 5: Confusing Zero Trust with Zero Friction

Zero Trust requires verifying every access request based on identity, device health, location, and other signals. Done well, this is largely invisible to users. Done badly, it creates significant friction that drives users to find workarounds — and each workaround is a hole in the architecture.

What to do instead: Design for user experience from the start. Risk-based authentication reduces friction for low-risk, routine access while maintaining strong controls for high-risk scenarios.

Mistake 6: Skipping the Pilot Phase

Zero Trust programmes that attempt to deploy at enterprise scale from the start almost always run into problems. The architecture is more complex than anticipated, application dependencies weren't fully mapped, or the identity governance work takes longer than expected.

What to do instead: Start with a well-defined pilot. Choose a business unit or application set that's representative of your broader environment, but small enough to move quickly. Use the pilot to validate your architecture before scaling.

Mistake 7: Treating Zero Trust as a Project, Not a Programme

Zero Trust is not a project with a defined end state. It's an ongoing programme of continuous improvement. The threat landscape evolves, the regulatory environment changes, new applications and workloads are deployed, and the architecture needs to evolve with them.

What to do instead: Establish Zero Trust as a programme with ongoing governance, regular maturity assessments, and a continuous improvement cycle.

Discuss your Zero Trust programme with a Cybinity security architect

Zero TrustNISTSecurity ArchitectureEnterprise SecurityCISA