Transparency is a central requirement in personal data governance, yet users are frequently exposed to legal disclosures they do not understand. This paper conceptualises transparency as a dual construct: a structural dimension, covering the formal conditions of information provision, and a cognitive dimension, addressing observable conditions for comprehensibility. Within the GDPR and the WP260 guidelines, a conceptual evaluation model is operationalised through a generic grid of 24 observable criteria organised around these two dimensions. It is applied to an exploratory audit of 20 e-commerce websites across five sectors in France, conducted between January and March 2026. The audit yields an overall transparency score of 90.7%, with a structural score of 95.4% and a cognitive score of 86.0%. This gap indicates that legal compliance alone is insufficient to guarantee genuine intelligibility. The framework is designed to support reproducible assessment through explicit criteria and scoring rules, and to be transferable beyond e-commerce to any digital environment where personal data are collected and disclosed.