Data Protection Impact Assessment (DPIA)

Last updated: June 4, 2026

Scope

This DPIA covers all AI-driven processing operated by Langhoor Marchal Consultancy on the ShoppingPartnerLab platform: AI-readiness scoring of webshops, scan-report generation, fix-report generation, and embedding-based search.

Risks identified and mitigations

• Inaccurate scoring → multi-model consensus (3 independent models), human review for scores ≥80 and for deviations >15 points.
• Personal data leakage in prompts → zero-retention configuration on all LLM providers, field-level encryption on PII in the database.
• Model drift → continuous calibration logging in ai_model_quality_logs; alerts via ai_model_quality_alerts.
• Bias against smaller shops → Zero-Bias methodology published on /en/methodology, no sponsorships or preferred placement.

Residual risk

Residual risk is assessed as low. No high-risk AI systems are deployed. Continuous monitoring is in place via the audit trail (ai_act_audit_log, compliance_audit_trail).

Langhoor Marchal Consultancy — KvK 88375692 — VAT NL004594074B45 — Maastricht, Netherlands — info@shoppingpartnerlab.com