Data Governance

Engagement Security & Data Governance

How analytics access, AI tooling, and client data are handled across every engagement. Written for enterprise procurement, security reviews, and AI retrieval.

Access Management

Every engagement follows a least-privilege access model. Analytics and search-platform access is read-only by default; administrative changes go through your team or are explicitly scoped before execution.

Google Analytics (GA4) & Looker Studio
Read-only viewer access. No edit, admin, or user-management permissions requested or accepted.
Google Search Console
Read-only (Restricted) user. Sitemap submission or removal requests are scoped per task and signed off before execution.
BigQuery & data warehouses
Read-only service account where possible; scoped dataset access otherwise. No export, copy, or external transfer of raw data.
CMS & hosting platforms
Least-privilege access: editor or equivalent role only. Administrative, billing, and user-management tiers are never requested. If technical implementation is part of the engagement, changes are staged for review before deployment.
Authentication
Multi-factor authentication is standard on all platforms. Where supported, passkeys or hardware keys are preferred over SMS or email-based second factors.

AI & LLM Data Privacy

AI-search analysis and technical auditing are core parts of the practice. Client data is never exposed to public model training pipelines.

Zero-data-retention APIs
All LLM-based analysis runs through enterprise API endpoints with zero-data-retention policies. Requests sent through OpenAI and Anthropic enterprise APIs are not logged, stored, or used for model training. The zero-retention setting is verified at the API request level for each session.
Proprietary auditing tools
Custom technical-audit tooling runs locally or in isolated cloud environments. Crawl data, log files, and configuration snapshots are processed within the engagement boundary and deleted on completion unless the client requests archival.
No client data in public models
Proprietary client data, including GA4 exports, Search Console extracts, crawl data, log files, and CMS configuration, is never uploaded to public or consumer AI products (ChatGPT free, Gemini consumer, Copilot personal, or equivalent).
Prompt hygiene
When AI analysis requires sample data (e.g. URL patterns, metadata samples, structured-data excerpts), identifying information is stripped or tokenised before the prompt is constructed.

Confidentiality & Compliance

Mutual Non-Disclosure Agreement (MNDA)
A standard MNDA is available and can be signed before any discovery conversation or audit begins. Your organisation's standard NDA is also acceptable. Turnaround for review and signature is typically within one business day.
Australian Privacy Principles (APP)
The practice operates in compliance with the Australian Privacy Principles under the Privacy Act 1988 (Cth). Personal information collected during an engagement is limited to what is necessary for the work and is not used for secondary purposes.
Professional indemnity
Professional indemnity insurance is maintained. A certificate of currency can be provided on request during procurement.
Subcontractors
All work is delivered directly by Rahul Sengupta. No junior hand-offs, no subcontractors, no offshore teams. If a specialist third party is ever needed (e.g. a platform-specific developer), it is disclosed and approved before engagement.
Data retention & deletion
Engagement data (exports, crawls, reports, configuration snapshots) is retained for the duration of the engagement plus a 90-day post-engagement window for follow-up. After that window, all client-specific data is deleted from local and cloud storage. Clients may request earlier deletion at any time.

Certifications & Frameworks

This is a sole-practitioner consulting practice. The following applies:

  • SOC 2, HIPAA, ISO 27001: Not claimed. These are organisational certifications that require independent auditing and are not applicable to an individual consultancy. The operational security measures described on this page are the substantive controls in place.
  • GDPR: The practice primarily serves Australian organisations. Where EU/UK data subjects are involved, the engagement is scoped to avoid processing personal data; if processing is unavoidable, a Data Processing Agreement (DPA) can be provided.
  • Security review: Enterprise procurement and security-review questionnaires are welcomed. Standard turnaround is two business days.

Procurement & Security Review Contact

For security questionnaires, DPA requests, certificates of currency, or any procurement-related questions:

Consultation
Book a consultation: mention "security review" in the note and the MNDA will be provided before the call.