Risk appetite is often presented as the point at which strategy and risk meet. It is described as a statement of intent, a boundary within which the organisation is prepared to operate, and a guide to decision-making under uncertainty. In principle, it is one of the most important artefacts in governance. In practice, it is frequently one of the least consequential. This is because risk appetite, as expressed, is rarely the same as risk appetite as it is lived in practice.
Formal statements are carefully constructed. They balance ambition with caution, growth with resilience, and opportunity with control. They are designed to be defensible to regulators and reassuring to boards, and in many cases they are entirely sincere. Yet they exist alongside a second, more powerful form of appetite that is rarely articulated. This is the behavioural appetite revealed through decisions, incentives, and trade-offs made under pressure. Over time, this divergence becomes the behavioural expression of a deeper issue: the organisation’s narrative is no longer being enacted in the way it is described.
This divergence does not usually present as failure. It presents as performance. Decisions that stretch stated appetite are framed as exceptions, justified by context, or absorbed into tolerances. Over time, exceptions accumulate and become normalised. The formal statement remains intact, while the lived boundary shifts, and behaviour moves further from the narrative it is meant to reflect.
Traditional risk management attempts to manage this through limits, metrics, and escalation protocols. These mechanisms are necessary, but they operate within the same frame. They assume that appetite has been clearly defined and that deviations can be objectively identified and corrected. In reality, many of the most consequential decisions sit in ambiguous territory, where interpretation matters more than measurement and where alignment cannot be reduced to compliance with thresholds.
The connection to narrative is direct. The stated appetite reflects the official story of the organisation. The lived appetite reflects how that story is enacted under pressure. Where the two diverge, narrative drift is already underway. Appetite becomes a stabilising artefact, preserving coherence in representation while behaviour adjusts in practice. What appears aligned at the level of reporting may already be misaligned at the level of decision-making.
Risk Intelligence approaches appetite from the opposite direction. It does not begin with the statement and attempt to enforce it. It begins with behaviour and asks what that behaviour reveals about the organisation’s true priorities. Appetite is not something to be asserted and monitored in isolation. It is something to be inferred from patterns of action, particularly when trade-offs are being made under pressure.
This requires close attention to incentives and responses. What is rewarded when outcomes are achieved through boundary stretching? What is tolerated when results are positive but methods are questionable? How are concerns treated when they challenge successful decisions? Appetite is encoded in these signals far more reliably than in formal articulation. Over time, they reveal whether the organisation’s declared posture is being sustained or quietly adapted.
Artificial intelligence introduces an additional layer of complexity. Systems designed to optimise within defined parameters will tend to explore the edges of those parameters. Where appetite is expressed in quantitative terms, AI will push systematically against those limits. Without careful governance, this can result in progressive stretching that remains technically compliant while becoming strategically misaligned. The organisation appears disciplined, even as its exposure increases.
If appetite is behavioural rather than declarative, then systems built to monitor limits will only ever capture part of it. The more consequential question is whether decisions, incentives, and responses remain aligned with what is stated. That alignment cannot be enforced through metrics alone. It requires visibility of how judgement is exercised in practice, particularly in situations where interpretation matters most. Environments that make this visible are not supplementary. They are foundational. RiskMetrica is built to expose that alignment, or the lack of it, in a way that cannot be reduced to thresholds or tolerated as isolated deviation.
Within such an environment, appetite is not reduced to thresholds and tolerances. It is observed through patterns of decision-making over time. Signals that indicate boundary stretching are connected, revisited, and interpreted rather than dismissed as isolated exceptions. The relationship between stated intent and lived behaviour becomes visible, and therefore open to challenge in a way that static artefacts cannot support.
For boards, this creates a more demanding form of oversight. The question is no longer whether appetite has been defined, approved, and cascaded, but whether it is being enacted. That requires engagement with decisions as they occur, not only with summaries after the fact. It also requires a willingness to confront misalignment even when outcomes appear favourable, and to recognise that apparent success can conceal underlying drift.
The question is not whether appetite has been defined, but whether it is being lived.
Richard Anderson
Founder, AndersonRisk
This essay was first published on 14 May 2026 on my Substack.