// Banking · Automated credit

How can a bank replay an automated credit decision?

By preserving the complete institutional decision—not only the score, model output or adverse-action reason.

A replayable credit decision connects the customer application to the policy and authority in force at that moment, the data and model versions used, the rules and judgement applied, the decision executed, the customer outcome and any subsequent correction.

// 01

Why a model explanation is insufficient

A model may explain which variables influenced a score. That does not establish whether the correct lending policy was encoded, whether an override was available, whether the decision fell within delegated authority or whether the customer received the outcome the institution intended.

Constructed example: a bank changes its strategy to protect viable small businesses during a downturn. The policy document changes, but the automated affordability rules do not. Each decline is technically consistent with the live rule. The portfolio outcome has nevertheless drifted from the authorised intent.

// 02

The evidence required

Authority

Who or what was permitted to approve, decline, price or escalate this application?

State

Which policy, data, feature, model, rule and threshold versions were active?

Judgement

Where could professional judgement enter, and was an override practically available?

// 03

Replay one decision

Start with the outcome received by the customer.
Recover the actual decision and its identifier.
Reconstruct the judgement, rules and model state applied.
Reconnect them to the intent, strategy and purpose that authorised the decision.
Establish whether feedback detected or corrected any divergence.

This is the Decision Integrity Chain™ applied to credit: Purpose → Strategy → Intent → Rules → Judgment → Decision → Outcome → Feedback.

// Questions people ask

Common questions

Is an adverse-action reason a replay?

No. It explains a stated reason for the outcome; it does not reconstruct the entire authority and decision chain.

Must the identical model output be reproduced?

No. Replayability does not claim determinism. It requires sufficient evidence to establish how the authorised decision became the actual outcome.

What is the first practical test?

Select one contested decline and ask whether it can be reconstructed without assembling a narrative from multiple teams.

// The practical test

Test one institutional decision

The ten-day Decision Drift Audit™ maps one material decision across all eight layers, assesses replayability and authority boundaries, and delivers one prioritised board finding.

Test one decision →Canonical definitions →Decision Integrity Chain™ →