// Board control · Decision drift

How can a board detect decision drift?

By testing whether actual outcomes still connect to authorised intent—not by asking whether every component is operating as designed.

Decision drift is the gradual separation of an executed action from the decision an institution authorised, occurring across the joins of the chain without any individual component failing visibly.

// 01

Why dashboards miss it

Dashboards report performance against the measures they were designed to observe. Drift often begins in translation: strategy becomes policy, policy becomes rules, rules become system logic and system outputs become operational decisions. Each component can remain within tolerance while the overall outcome moves away from the board’s intent.

Constructed banking example: the board changes its appetite toward liquidity resilience. Treasury incentives, deposit classifications and automated pricing rules change at different speeds. No threshold is breached, yet the institution gradually behaves as though yield remains the dominant objective.

// 02

Five indicators

The same policy produces materially different outcomes across channels.
Overrides exist formally but are rarely or never exercised.
A board decision changed, but no one can identify every rule and system affected.
Reconstruction of a contested decision takes weeks and remains incomplete.
Feedback reports outcomes but cannot alter the decision chain.

// 03

The board test

Select one consequential decision from the last 90 days. Ask management to reconstruct Purpose → Strategy → Intent → Rules → Judgment → Decision → Outcome → Feedback. Do not ask each function whether its control worked. Ask where the complete decision changed, who held authority at that join and whether the change was explicitly authorised.

That is the difference between component assurance and decision integrity.

// Questions people ask

Common questions

Is decision drift the same as model drift?

No. Model drift concerns changes in model behaviour or data relationships. Decision drift can occur even when every model performs exactly as validated.

Does automation create decision drift?

No. It reduces the time available to detect and correct drift, and can repeat it at institutional scale.

What is the first board action?

Ask for one end-to-end replay of a consequential decision rather than another aggregate control dashboard.

// 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™ →