Begin with the resource, not the product
NIST SP 800-207 describes zero trust as a move away from implicit trust based on network location. For an enterprise, the first useful question is which resources must remain available and protected: applications, data stores, management interfaces, and workflows. Record each resource owner, business impact, data sensitivity, and the users and services that need access. This creates a scope that a security team can actually review.
Map the path of access
For every priority resource, list human and service identities, their devices, authentication methods, current permissions, and the network or cloud path they use. Include employees, contractors, partners and administrators. Compare the permissions granted with the work each identity must perform. Unknown owners, shared accounts, stale privileges and unobserved service connections become concrete remediation tasks rather than vague zero trust objectives.
Test the inventory against a purchasing workflow
Consider a hypothetical planning example, not a customer case: employees submit purchase requests from the office or remotely, an outsourced finance team reviews them, and the approval application calls an inventory interface. A list limited to servers and employee accounts may miss contractor access expiry, interface identities and attachment storage. Ask the business owner to walk through a normal request: who initiates and approves it, where attachments sit, where the application obtains data, and who handles access failures. Map each step to its resources and accessing identities.
Then examine exceptional paths: a role change, an expired outsourcing contract, a lost device or an interface key rotation. The immediate goal is to identify exceptions without owners, deadlines or alternatives, rather than block them all. This workflow is a starting point, not evidence of an enterprise-wide inventory.
Make the inventory useful for an access decision
Record the resource, business purpose, environment, owner, user categories, identity source, device requirements, required actions, access approver, review date and log location. Break 'finance access' into actions such as viewing, exporting and approving because their effects differ. Store references to controlled systems, not passwords, private keys or complete customer records.
Label unknowns and record whether evidence came from configuration, logs or interviews. An observed connection proves that a path existed during observation; it does not establish continuing business need or approval. The business owner confirms necessity, the technical owner confirms implementation, and the security owner assesses gaps. A scan cannot replace those judgments.
Separate human and system identities
Employee access can follow hiring, transfers and departures. Service accounts, scheduled jobs and application interfaces need technical owners and rotation plans. Before removing a shared account, identify dependent workflows and plan attributable alternatives so overnight jobs keep working. List administrative entry points separately from normal business access to give privileged actions a clear review scope.
Partners also need expiry reviews and revocation conditions. Contract termination, task completion and owner changes can trigger checks instead of leaving permissions indefinitely. For legacy systems that cannot be changed immediately, document constraints, compensating controls and a reassessment date. Calling those systems fully migrated would conceal remaining risk.
Observe before enforcing the approved scope
The business owner defines legitimate work; identity staff manage accounts and permissions; endpoint staff verify device state; application staff explain interfaces; and security staff assemble decisions and logs. One person may hold multiple roles in a small team, but accountability must be explicit. Preserve existing policies, support contacts and rollback steps, and agree which incorrect denials require a pause.
Compare observed activity and business interviews with proposed policies before enabling them for representative users and devices. Test permitted and prohibited actions: an authorized approver succeeds, a revoked partner cannot enter, and a device that fails requirements receives understandable instructions. Blocking every request is not proof of success.
Use acceptance evidence to decide whether to expand
For each test, record identity category, device state, resource, action, expected and actual outcomes, and log time. Acceptance includes correct allowance and denial, support staff being able to explain decisions, users completing work, and exceptions following the response plan. Business stakeholders should agree experience thresholds in advance rather than borrow another enterprise's figures.
Assign owners and deadlines to missing interfaces, incorrect denials, excessive permissions and logging gaps. Expand only after these issues are resolved or explicitly accepted. Track the share of resources with confirmed owners, overdue reviews and remaining exceptions, keeping the denominator consistent so a shrinking scope cannot manufacture improvement.
Maintain the inventory through ordinary change control
New applications, acquisitions, supplier replacements and identity-platform changes alter access relationships. Require inventory updates as part of their handover and sample records against production. Zero trust does not remove the need for backups, patching, incident response or network segmentation. It requires evidence-based, traceable access decisions that can be reviewed over time. These implementation steps are JOTOglobal recommendations based on public principles, not a NIST certification of a particular enterprise.
Turn inventory into a staged plan
NIST SP 1800-35 documents 19 example implementations using different approaches. These are examples, not a universal shopping list. A practical pilot protects one high-value workflow, defines measurable access decisions and logs, tests legitimate user experience, and records exceptions. Expand only after the team can explain why access was allowed or denied and how the resource will be recovered if a control fails. The inventory should be refreshed as assets and business relationships change.
Sources
Zero Trust Architecture (SP 800-207)
https://csrc.nist.gov/pubs/sp/800/207/final
Implementing a Zero Trust Architecture (SP 1800-35)
https://csrc.nist.gov/pubs/sp/1800/35/final
