Skip to content
ConsultEvo

How to Evaluate GDPR-Supporting Software Across Your Business

Evaluate GDPR-supporting software by testing whether the exact product, edition, tenant, region, plan and configuration can support your documented data flows, purposes, lawful bases, security, retention and rights processes. A meeting recording, transcript and AI summary may live in different places and need different access and deletion rules, even when they come from one call.

Start with the data and business purpose, not a vendor’s “GDPR compliant” label. A vendor’s data processing agreement (DPA), privacy page or trust-center material is evidence to review. It does not prove that your organization’s use is compliant. This guide is an operational evaluation framework, not a legal determination. A lawful basis is required, but the appropriate basis and any consent requirements depend on purpose, context, jurisdiction and applicable national law. For uncertain decisions, consult qualified counsel.

The goal is not to find a universally compliant product. It is to determine whether a specific configured system can support a defined processing activity, what evidence proves that, and which responsibilities remain with your organization.

The short answer: evaluate controls in your actual data flow

Include meeting software and every CRM, marketing or sales system that receives meeting-related personal data. For each flow, identify what is collected, why it is processed, where it goes, who can access it, how long it is kept and how an individual’s rights can be handled. The European Data Protection Board’s overview of lawful bases is a useful starting point, but it is not a case-specific legal assessment.

A software control is useful only when its documented scope, configuration, data category and operational owner match the processing you actually intend to do.

Map the data before comparing vendors

Make one inventory row for each data category and processing purpose, not one row per vendor. A video recording and its transcript may have different storage locations, access groups and deletion paths. Include collection, processing, storage, access, sharing, transfer, export, retention and deletion in the trace.

  • Meeting records: participant details, attendance, chat, shared files and calendar information.
  • Meeting artifacts: audio or video recordings, transcripts, AI summaries and notes.
  • Downstream copies: CRM activities, marketing events, sales records, integration copies and exported files.

For each row, record the purpose, proposed lawful basis, source of truth, system owner, recipients, retention rule and evidence location. Do not collapse “EU data residency” into one answer about location. Storage, live media processing, backups, analytics, billing, AI processing, support access and subprocessors can be separate questions. The GDPR text provides the legal framework, but your organization must assess its actual transfer circumstances.

01Name the purpose and dataThe process owner identifies each category, such as a recording, transcript or CRM note, and documents its purpose and proposed lawful basis.
02Trace each artifactThe system owner records where it is collected, processed, stored, shared, exported and deleted, including downstream CRM or marketing copies.
03Verify evidence and scopeProcurement and privacy reviewers check current product documentation, DPA, subprocessors, transfer terms and whether each control covers the specific artifact and tenant.
04Test configuration and exceptionsThe platform administrator tests access, retention, export and deletion settings. The privacy owner resolves lawful-basis or transfer questions.
05Record the decision and ownerProcurement logs each gap as acceptable, configuration work, contract or process work, or a reason not to use the product for that flow.

Use evidence-based criteria, not a feature-count score

For each requirement, record the control, evidence and scope, plan or configuration dependencies, and person responsible for operating it. Mark unknowns for follow-up rather than treating them as supported. Ask vendors which data categories a regional option covers, and whether it includes processing and support access as well as storage.

Control What to verify Evidence and owner
Processing and transfers Data categories, locations, subprocessors, DPA and applicable transfer mechanism Vendor documents; procurement and privacy review
Security and access Encryption mode, administrator permissions, auditability and feature limits Product documentation; platform administrator
Retention and deletion Rules for each artifact, exceptions, holds and deletion timing Configuration test; system owner
Rights and integrations Export and deletion scope, connected systems and downstream copies Test request; privacy operations and system owners

Official documentation shows why scope matters. Zoom documents data-subject-request settings and retention behavior that differs by data category. Microsoft documents Teams compliance functions through Purview, including retention and eDiscovery. Teams content can sit across different Microsoft 365 workloads, and holds can affect deletion timing. Google documents encryption and recording storage behavior for Meet. Webex documents regional residency with exceptions for some data categories. These sources evidence particular controls, not universal availability or activation in every deployment.

Decision point

A stronger encryption mode can limit cloud-dependent recording, transcription or AI functions. Webex documents feature restrictions for some end-to-end encrypted meeting modes, while Google documents different behavior across meeting types and qualifying encryption configurations. Test the required security mode and business function together.

Use a status such as “documented for this scope,” “plan or tenant dependent,” “administrator configuration required,” “limited by another security mode,” or “not verified.” Record the documentation date and set a review date because vendor terms and product behavior can change.

Govern recordings as separate data assets

Before enabling recording, transcription or AI summarization, determine whether each artifact is necessary for the stated purpose, where it will be processed and stored, who can access it, and when it should be reviewed or deleted. A recording notification is not proof of consent. Recording and sharing a meeting create data-protection obligations. The ICO’s guidance on recording and sharing online meetings advises organizations to consider and document the applicable lawful basis.

Illustrative recording decision gate

Input includes a tenant, meeting identifier, source system, requested artifacts, purpose code, notice status, basis-review status, retention category, access group and proposed storage location. The meeting organizer submits the request. A privacy or legal reviewer owns uncertain basis decisions, and the platform administrator checks technical settings.

AI may extract a candidate purpose or identify that a transcript or summary was requested. It must not determine whether processing is legally permitted or whether consent exists. The proposed gate below is an operating design, not an official vendor workflow or verified cross-platform integration.

{
  "tenant_id": "tenant_19",
  "source_system": "meeting_platform",
  "meeting_id": "mtg_2026_1042",
  "request_id": "rec_req_771",
  "purpose_code": "customer_support_review",
  "artifacts_requested": [
    "recording",
    "ai_summary"
  ],
  "participant_notice_status": "pending",
  "lawful_basis_review_status": "pending",
  "retention_category": "support_case",
  "access_group": "support_managers",
  "decision": "hold",
  "review_owner": "privacy_operations"
}

Before allowing the request, validate the approved purpose and basis review, notice status, artifact locations, access group, retention category and security-mode compatibility. A failed notice or missing review sends the request back to its named owner rather than silently enabling capture. If the meeting produces multiple recordings, transcript versions or regenerated summaries, each asset needs its own source asset identifier and version. A meeting ID alone is not a safe asset key.

Test rights requests across systems, not just the CRM

A rights-request process should search every relevant system and record what happened in each one. A request after a recorded sales call might involve the meeting participant record, recording, transcript, summary, chat, calendar invitation, CRM activity, marketing audience and sales tool. A deletion action in one system should not be assumed to remove copies elsewhere.

Use one parent request and a child task for each system or audience. Verify the requester, search known identifiers and aliases, record export artifacts and actions, check holds, and require human approval before irreversible deletion. Distinguish permanent deletion from suppression, anonymization, archiving or deletion delayed by a hold. Mailchimp documents that contact exports are audience-specific, so every relevant audience must be considered.

Check each request before closing it
  • Requester identity and relevant identifiers have been verified.
  • Meeting artifacts, CRM records, connected systems and relevant marketing audiences have been searched.
  • System-specific exports and actions are recorded. One audience export is not assumed to cover another.
  • Holds and the meaning of deletion, suppression or anonymization have been checked.
  • A named privacy operations owner has accepted exceptions and unresolved tasks.

To prevent retries from creating duplicate tasks, use a database-enforced unique key such as request ID plus source system plus export type, or an equivalent transactional upsert. A read-then-create check alone can race when jobs run concurrently. For an export row, the grain is one export artifact for one request, source system and export type. If a system generates a replacement file, store a new artifact version rather than overwriting the evidence record.

Keep consent evidence tied to purpose and notice

Meeting attendance does not, by itself, establish permission for marketing. Evaluate whether the CRM or marketing system can preserve the purpose, lawful basis, consent status, timestamp, notice or form version, source record, and withdrawal or suppression state. If the marketing purpose differs from the meeting purpose, assess it separately.

HubSpot documents consent and lawful-basis concepts in its privacy glossary. Mailchimp documents GDPR form fields and consent text, while cautioning that enabling those fields alone does not make an organization compliant. Its guidance also explains how to collect consent and preserve form information. For imported contacts, retain the underlying consent evidence. A vendor form setting is not a substitute for it.

Illustrative marketing activation rule

A proposed CRM or marketing eligibility check can allow activation only when evidence is present for the exact purpose, the notice version is known, and no later withdrawal or suppression exists. Use deterministic rules for these checks. AI classification should not change consent status. Missing or conflicting records go to a named marketing operations reviewer, with privacy review for lawful-basis or purpose questions. Do not activate a contact simply because an attendance event arrived.

{
  "subject_id": "subj_427",
  "purpose_code": "product_news",
  "source_record_id": "crm_contact_915",
  "consent_event_id": null,
  "consent_status": "unknown",
  "notice_version": null,
  "withdrawal_timestamp": null,
  "activation_decision": "block",
  "reason_code": "missing_evidence"
}

These fields and the gate are a proposed design, not a vendor-published schema or verified connector. A consent-evidence row should represent one consent event for one subject, purpose, notice version and event time. Do not use contact ID alone as the key, because one person can provide or withdraw consent for multiple purposes and form versions. Save the evidence reference and decision in the designated system of record, and send unresolved conflicts to a review queue. Organizations assessing fragmented contact ownership or inconsistent consent evidence may also need to review their CRM systems.

Separate documented controls from proposed operating designs

The sources support individual vendor capabilities, not a complete cross-vendor automation. Treat the following as an evaluation pattern that your technical team must implement and test:

Trigger AI job Validation Action and fallback
Organizer requests recording, transcript or summary Suggest purpose or identify requested artifacts Check notice, basis review, storage, access, retention and encryption-mode compatibility Hold until approved; route legal questions to privacy and configuration failures to the administrator
Verified access or deletion request Summarize the request for a reviewer Verify identity, create one task per system or audience, check holds and record outcomes Export or act in each system; require human approval for irreversible deletion
Meeting-related contact considered for marketing No consent decision; optional advisory purpose classification Require exact-purpose evidence, notice version and no later withdrawal Allow only when evidence passes; otherwise block or send to marketing operations review
Vendor or region changes Summarize changed documentation for review Recheck DPA, subprocessors, locations, transfer mechanism, plan and feature scope Update the assessment or pause the affected flow until the owner accepts the change

For concurrent automation, use database-enforced uniqueness or a transactional upsert at the correct grain. Examples include tenant plus source system plus source asset ID plus asset type and version for recording assets; subject plus purpose plus notice version plus consent event timestamp for consent evidence; and request ID plus source system plus export type for rights-request exports. These are proposed implementation keys, not vendor-published fields.

Turn findings into a vendor decision and operating plan

Compare products against the flows you intend to run, not a generic feature checklist. Request current, product- and tenant-specific evidence for DPA terms, subprocessors, transfer mechanisms, region scope, retention and rights workflows. For international transfers, identify whether a restricted transfer occurs and which mechanism applies. Standard Contractual Clauses are one possible mechanism, not an automatic requirement for every international participant or transfer.

Name an owner for administrator configuration, privacy decisions, CRM and marketing data quality, request fulfillment and vendor review. If HubSpot is under consideration, review the relevant product settings and subscription directly. Organizations assessing its configuration may find HubSpot systems relevant. Neither a product nor a service guarantees compliance.

For every gap, make an explicit decision: accept it for the intended use, resolve it through configuration, address it through contract or process, or disqualify the product for that flow. Revisit the evidence when plans, tenants, regions, vendor terms or processing purposes change.

What to verify before procurement sign-off

  • Each personal-data category and purpose has a documented owner, source of truth and retention rule.
  • Recordings, transcripts, summaries and downstream copies have been evaluated as separate assets.
  • Vendor evidence covers the exact product, plan, tenant, region and configuration being procured.
  • Storage, live processing, support access, backups, analytics, AI processing and transfers have not been treated as one location question.
  • A test rights request and marketing activation gate have named owners, recorded outcomes and human exception paths.
  • Proposed automation has been tested for retries, concurrent runs, replacement artifacts and conflicting records.

The vendor facts cited here are selected examples from official documentation, not a complete feature audit of every CRM, marketing or sales platform. Confirm current scope and contract terms for your deployment, and involve qualified counsel where the legal basis, transfer mechanism or recording requirements are uncertain.