Talk to a specialist

PCI DSS · Practical implementation guide

PCI DSS encryption, key management and data in transit

Encryption protects payment data only when the organisation understands where protection begins and ends, who controls the keys and how failures are handled. Build a cryptographic inventory that connects data flows, storage, certificates, keys and administrators. Then verify the implementation rather than relying on an enabled setting or a product’s marketing description.

In this guide

Identify the protection goal for each use

PCI DSS v4.0.1 addresses stored account-data protection and strong cryptography for cardholder data transmitted over open, public networks. The detailed requirements and applicability notes determine the formal obligation. Treat this article’s design workflow as practical implementation guidance alongside the complete standard. Reference: PCI SSC current standard and testing procedures.

Separate storage encryption, transport protection and key protection. Storage controls address retained information; transport controls address a communication path; key management controls address the material and authority that make encryption effective. A system can use strong transport encryption while logging unprotected data at an endpoint, or encrypt a database while exposing its decryption credentials broadly.

Write the threat and boundary for each use. Are you protecting a backup from unauthorised reading, a browser connection from interception, or a key from extraction? Identify the trusted endpoints and what happens to data after decryption. This prevents a broad phrase such as end-to-end encrypted from concealing important intermediate processing.

Challenge unnecessary data before selecting cryptography. If a business process can operate with a provider reference or reduced dataset, avoiding retention may simplify both security and operations. Encryption should protect justified information, not become a reason to keep every field indefinitely.

Create a cryptographic inventory linked to services

Record the service, data protected, mechanism, algorithm or protocol configuration, key or certificate reference, owner and dependencies. Do not put secret key material into the inventory. The record should identify where authorised administrators can find controlled evidence and which system manages the secret.

Include load balancers, gateways, application connections, database encryption, backups, file transfers, remote administration and provider interfaces. Examine regional and recovery variants. A production endpoint may use current settings while a dormant recovery endpoint retains an older configuration that becomes active during an outage.

Map termination points. If a browser connection ends at an edge service, determine how data travels to the next component and who administers both ends. Record what is visible in logs and diagnostic captures. Security depends on the complete path, not only the first connection visible to the customer.

Identify ownership gaps. Certificates obtained by a former employee or keys controlled by an undocumented automation account create continuity problems. Assign a primary and backup owner, renewal or review process and escalation route. Inventory quality should support both security assurance and prevention of avoidable service interruption.

Evaluate TLS as an implementation, not a version label

PCI SSC FAQ 1491 states that PCI DSS does not prescribe one Transport Layer Security (TLS) version. Its July 2026 clarification excludes all SSL versions and, at a minimum, TLS 1.0 and 1.1 from strong cryptography. It also requires attention to secure algorithms, settings and implementation. Therefore, a version number alone is insufficient evidence of a secure configuration. Reference: PCI SSC TLS clarification.

Review the protocols and cipher suites actually offered, certificate validation, trust configuration and client behaviour. Consider every supported endpoint and relevant client path. A public website test may not examine an internal application-to-provider connection or a batch-transfer endpoint. Record the coverage of each test and resolve unreachable or untested paths.

Test failure behaviour. A client should not silently continue in an unsafe mode because a certificate is invalid or a secure connection fails. Application teams should understand how errors are surfaced and how operations will respond. Availability pressure can lead to dangerous temporary workarounds if the legitimate recovery procedure is unclear.

Monitor cryptographic changes over time. Libraries, vendor defaults and threats evolve. Maintain a review process that identifies affected services and supports planned configuration updates. Do not let a successful assessment from a previous year become the only basis for retaining an old cryptographic implementation.

Design the key lifecycle around authority and recovery

Document how keys are generated, stored, distributed, used, changed, retired and destroyed. For each stage, identify the people and systems with authority. The design should minimise unnecessary access to key material and make sensitive operations reviewable. Key management is an operating process, not simply a choice of encryption algorithm.

Separate access to encrypted data from authority over decryption where practical and appropriate to the architecture. Examine cloud role combinations, service identities and administrative consoles. An individual who cannot read the database directly may still grant themselves decryption rights through another role. Review effective permissions rather than individual permission names in isolation.

Plan rotation with application behaviour in mind. Identify how existing data remains readable, how new writes select the active key and how failures are detected. Test migration and rollback in a controlled environment. A rotation that is recorded as complete while old dependencies remain undocumented can create both security and availability risks.

Design recovery carefully. Lost keys can make legitimate data unavailable; uncontrolled backup copies can weaken protection. Record the recovery mechanism, access approvals and verification process. Test recovery without exporting secrets into informal notes or general-purpose collaboration systems. The evidence should demonstrate the process while preserving the confidentiality of the material itself.

Understand provider-managed encryption

Managed services can reduce operational work, but responsibility remains divided. Ask what the provider encrypts, which keys are used, who can administer them and which settings the customer controls. Verify whether the service’s compliance evidence covers the relevant function and whether customer configuration affects its security.

Do not assume encrypted account data is automatically outside PCI DSS consideration. PCI SSC explains that encrypted cardholder data and the systems managing encryption, decryption and keys can remain relevant to scope. The precise result depends on the environment and relationships. Reference: PCI SSC encrypted-data scoping guidance.

Review provider access and support arrangements. Can the provider reveal data, restore a snapshot or alter a key policy? Can your own support staff request those actions? Record authorisation, notification and evidence expectations. A responsibility matrix should describe operational actions rather than merely assigning encryption to the provider.

Consider exit and portability before signing. If the service ends, how will data be transferred, retained or removed, and what happens to the keys? Test assumptions about export formats and decryption authority. A secure service can still create a serious business dependency if no one has planned the authorised exit path.

Hypothetical example: an insurer’s premium-payment portal

A fictional insurer operates a premium-payment portal behind a managed edge service. The browser connection is encrypted, and management assumes the full journey is covered. A data-flow review shows a separate connection from the edge to the application and a nightly file transfer used for reconciliation.

The team inventories all three paths and identifies the owners. It verifies the deployed transport configuration and certificate validation for each, then examines what information appears in application and transfer logs. Finance confirms that the reconciliation process needs a limited dataset, allowing unnecessary fields to be removed before transfer.

Operations establishes certificate ownership and renewal monitoring for the less visible batch endpoint. A controlled failure test confirms that an invalid certificate causes a clear operational alert rather than an insecure retry. The incident procedure explains who can restore service and which temporary changes require review.

The hypothetical lesson is that customer-visible encryption is only one part of a payment workflow. Back-office transfers and service-to-service connections deserve the same explicit ownership and evidence. The improvement comes from understanding the complete path, not from adding a stronger-sounding badge to the portal.

Hypothetical example: a platform rotating storage keys

A fictional payment-related platform uses a managed key service for encrypted storage. The engineering team schedules key rotation and records the new key identifier. Later, a recovery test finds that a historical dataset depends on an older key whose ownership and retention status are unclear.

The team maps datasets to key versions and identifies the application paths that read them. It distinguishes key activation, data migration and retirement as separate steps. The revised process requires evidence that relevant data remains readable through the authorised service and that obsolete keys are retired only after dependencies are resolved.

Access review also reveals that a broad cloud administration role can alter key policies. The platform narrows routine privileges and establishes a controlled route for sensitive changes. Logs of those changes are connected to review and incident processes. Recovery evidence contains identifiers and results, not secret material.

The example shows why rotation is a lifecycle exercise. Changing a key identifier does not automatically prove that old data, permissions and recovery arrangements have been addressed. Good evidence follows the data and authority through the transition.

Verify implementation and preserve useful evidence

Combine configuration review, controlled testing and operational records. Configuration shows intended settings; testing examines their effect; operational records show maintenance and exceptions. None should be overstated. A scan of one endpoint does not establish every connection, and a policy does not establish actual key permissions.

Retain a clear record of cryptographic review decisions. Include the affected services, observed configuration, changes required, owner and verification outcome. Link vulnerabilities or vendor changes to the systems that use the affected component. This allows a security advisory to become a targeted action rather than a generic message sent to every engineer.

Protect the evidence itself. Key inventories, architecture and access records can be sensitive even when they contain no secret material. Use appropriate access controls and sharing methods. Never include private keys or passwords in an assessment pack as proof that they exist. Demonstrate controlled processes without undermining them.

Common mistakes include assuming storage encryption covers application exports, treating key rotation as a checkbox, ignoring provider-managed settings and relying on one public TLS test for an entire service. Correct them by defining the protected data, trusted boundary and evidence coverage for each use.

Plan improvements without disrupting payments

Prioritise exposed weak configurations and unclear key authority, then sequence changes around dependencies. A protocol update may require client upgrades. A storage redesign may require data migration and recovery testing. Define a safe rollout, success criteria and rollback conditions before production changes.

Use a representative pilot where the environment is large or diverse. Include different client types and regional integrations. Record exceptions explicitly and give them owners. A successful pilot supports the rollout decision but does not prove that every untested legacy connection will behave identically.

Make the next cryptographic review specific: identify the data flows, termination points, key ownership and unresolved recovery questions. Explore PCI DSS consulting and contact CyberVantage360 to discuss assessment and remediation support. Confirm whether specialist key-management engineering and provider coordination are included in the engagement.

Document the migration decision

When replacing an older cryptographic mechanism, write down the assets affected, the reason for change, compatibility constraints and the evidence required for completion. Include both active and archived data. A project can update new transactions successfully while leaving old exports or recovery copies dependent on an obsolete mechanism.

Assign a person to reconcile the inventory after migration. Confirm that disabled endpoints are actually removed, old credentials are handled appropriately and monitoring reflects the new design. Retain the decision record so a future team understands why a legacy dependency was retained temporarily and what condition permits its retirement.

For cross-border services, verify each regional endpoint and provider integration. A shared application release may not update a locally managed gateway or certificate store. Central oversight should therefore compare deployment evidence, not simply collect regional declarations of completion. This is a practical programme recommendation for managing variation and avoiding false confidence during a global rollout.

Check who can change cryptographic policy

A key can be securely stored while its access policy remains broadly editable. Review the administrators who can grant decryption, alter trusted identities or change the application configuration that selects a key. Effective authority matters as much as direct access to secret material.

Trace one sensitive policy change through request, approval, deployment and logging. Confirm that the resulting permission matches the approved purpose and that a reviewer can identify the actor. Include automation that manages the policy, because its configuration may be controlled elsewhere. This examination helps distinguish a secure cryptographic service from a secure use of that service and provides a practical basis for narrowing privileges without disrupting legitimate application operation.

Frequently asked questions

Does PCI DSS mandate one TLS version?

No. Evaluate the complete implementation against strong-cryptography expectations and current PCI SSC guidance.

Does encryption remove all PCI DSS scope?

No. Data handling, decryption capabilities, key management and other security dependencies still need examination.

Should key material be included in evidence?

No. Use identifiers, controlled demonstrations and process records that establish the relevant facts without exposing secrets.

What should a key-rotation record demonstrate?

It should explain the authorised change, affected data and services, verification, remaining old-key dependencies and the planned retirement or recovery treatment.