Explainability (reason codes)

Explainability is the ability to show why a system reached a specific decision in terms a person can understand; reason codes are standardised statements of the main factors.

Part of the Learn hub and the customer data glossary.

What it means in practice

Explainability has a legal edge and a practical one. Several regimes ask for reasons:

• US credit: Regulation B requires a statement of reasons for adverse action that is specific and indicates the principal reasons; citing the creditor's internal standards or a failed credit score is insufficient (12 CFR 1002.9(b)(2)) 1.
• GDPR: the access right covers information about automated decision-making (Article 15(1)(h)), and Article 22(1) gives a right not to be subject to solely automated decisions with significant effects 2. In Case C-203/22 (27 February 2025) the Court of Justice held that the controller must explain the procedure and principles actually applied, concisely and intelligibly, not necessarily the full algorithm 3.
• EU AI Act: people affected by decisions based on Annex III systems can obtain "clear and meaningful explanations" of the AI system's role in the decision (Article 86(1)) 4.

Practically, record for every decision the inputs, the rule or model version, the top factors and a reason code mapped to plain language. Rules are explainable by construction. For machine-learning models, feature attributions give approximations and should be described honestly. Marketing choices are less regulated, but the same records speed up complaints and audits.

Example (retail bank): a credit-limit decision stores its inputs, model version and reason codes (for example high utilisation, short credit history), so the bank can give a specific statement of reasons rather than "did not meet our criteria".

Related terms

• Audit ledger: holds the decision records.
• Model lineage: ties decisions to model versions.
• High-risk AI system: where explanation rights are strongest.
• Real-time decisioning: where reasons are generated.
• Next best action: carries reason codes.

Where it matters

• Explainable AI for customer decisions: reason codes, lineage and what auditors ask.
• Explainable eligibility communications: audit trails for public benefits.

Back to the customer data glossary or the Learn hub.