When organisations forget what a good decision looks like

How changes in risk appetite can affect decision-making confidence, and what helps organisations rebuild it

Picture a routine decision that a capable, experienced staff member once made on their own in an afternoon. Today the same decision needs three sign-offs, a risk paper and a slot on the executive agenda. The decision is no harder than it was and the staff member is just as capable, yet nobody can say exactly when things changed or why. Nobody took the decision away from them. Somewhere along the way, they became afraid of getting it wrong and stopped trusting themselves to make it.

“Nobody took the decision away from them. Somewhere along the way, they became afraid of getting it wrong and stopped trusting themselves to make it.”

I have come to think this is one of the most underestimated risks in organisations, and particularly in regulators. When an organisation's appetite for risk shifts, whether through new legislation, a change of purpose or policy, or the memory of something that went wrong, people can lose their sense of what a good decision now looks like. Anxiety about making the wrong call sets in, and decisions that were once made with assurance start to travel upwards. Senior leaders become mired in operational detail, bottlenecks form and strategic work waits. And while that confidence can be lost quickly, it can take years to rebuild.

Over a career that has taken me through local and central government, telecommunications, energy, tertiary education, and professional regulation, I have seen this pattern repeatedly. It is rarely caused by people becoming less capable. More often, the organisation has changed what it is willing to accept without being clear about what that means for everyday decisions and has gradually lost the institutional memory that once gave people the confidence to act.

Artificial intelligence (AI) adds a new dimension. It could help organisations rebuild confidence by making the reasoning behind past decisions far easier to find, or it could deepen the problem by giving people one more reason not to exercise their own judgement. The real question is not whether AI can remember, but whether organisations can still learn, and whether their people will still have the confidence to decide.

Why risk appetite shifts

Most organisations treat risk appetite as something formal: boards approve risk frameworks, policies define acceptable levels of risk, and limits and authorities are documented. All of this matters, but in practice an organisation's appetite for risk is shaped as much by experience and circumstance as by any framework, and it moves more often than we tend to acknowledge.

New legislation can bring new duties, powers or expectations. A change in purpose or vision can redefine what the organisation exists to do and whom it serves. Policy changes can redraw the boundaries of acceptable practice, and new leadership, a restructure, an adverse review or a high-profile incident can each reset the tone almost overnight. These shifts are often necessary and right. The question is not whether they should happen, but whether organisations help their people adjust when they do, and too often they do not.

The difficulty is rarely the shift itself; it is the uncertainty that opens up behind it. When the ground moves, people are not always told clearly what has changed and, just as importantly, what has not. The reference points they once relied on no longer feel reliable, and unless the organisation's values are clear enough to steer by, it becomes genuinely hard to tell what a right or wrong decision looks like. Understandably, people become anxious about getting it wrong, and the easiest way to manage that anxiety is to pass the decision to someone more senior.

Memory plays a part too. People remember previous failures, investigations, regulatory reviews and complaints, and those experiences quietly shape thousands of decisions. When the people who lived through them move on, the lesson may survive in a report, but the understanding behind it often does not. A new generation of leaders inherits the caution without necessarily inheriting the reasoning, and without that reasoning it is very hard to know when caution is warranted and when it is not.

When confidence goes, everything escalates

Regulatory decision-making shows this especially clearly. Regulators' decisions can be challenged, appealed, reviewed and scrutinised in public, so when staff are unsure where the organisation now stands, the fear of making the wrong call is very real, and referring the decision upwards starts to feel like the only defensible option. Some decisions should rise - those that are novel, high-consequence or likely to set a precedent. The problem is when routine decisions follow them. Informal habits then harden into process: new approval steps appear, more matters require sign-off and escalation pathways expand. Each step seems sensible at the time, because nobody wants to be responsible for the next failure.

The same pattern appears well beyond regulators. Frontline staff defer to managers, managers defer to senior leaders, senior leaders seek further advice, and boards find themselves drawn into increasingly operational matters. No one believes they are being unreasonable; most people are acting in good faith, simply unsure what is now expected of them and anxious not to get it wrong. The problem is that escalation gradually becomes a substitute for judgement, and the organisation begins to lose confidence in its own ability to decide.

Busy, but less effective

As decisions move further from the point where the expertise actually sits, they rarely become proportionately better; they simply become slower. Queues and backlogs grow, work piles up around a handful of senior decision-makers, and the people affected by those decisions, whether applicants, practitioners, businesses or members of the public, wait longer for answers. The organisation becomes busy but less effective, and the cost of its caution is borne by the people it exists to serve.

The less visible cost falls on senior leadership. Executives who should be shaping strategy, building relationships and leading change find their diaries consumed by operational decisions their teams are perfectly capable of making. Strategic initiatives slip, often including the very changes that the new legislation, purpose or policy was meant to deliver. It is an irony I have seen more than once: a shift intended to move an organisation forward ends up anchoring its leaders in the detail.

Over time, capability erodes as well. People stop exercising judgement because they no longer feel sure enough to, and future leaders have fewer chances to develop it because fewer meaningful decisions are made at their level. The organisation gradually loses confidence in delegated authority altogether.

Why confidence is slow to return

Confidence can be lost in a single event, but rebuilding it usually takes years. Controls are easy to add and hard to remove, because nobody wants to be the person who took one away just before the next incident. Delegations may remain on paper yet go unused for years, and staff who have learned that escalation is the way to avoid blame will not change their behaviour because a memo tells them to. They take their cue from how the executive and the board respond when something goes wrong, and a single public rebuke can undo years of encouragement. They need a clear sense of what a good decision looks like in the new context, and they need to see reasonable decisions backed by their leaders, including when the outcome is not perfect, consistently over time.

This is where organisational values matter more than most organisations realise. When legislation, policy or risk appetite changes, the rules and precedents people relied on can suddenly feel unreliable, but the values that describe what the organisation stands for usually endure. Used well, they give people a way to judge a decision even when no rule covers it: a staff member who can ask whether a decision is fair, proportionate and consistent with the organisation's purpose has a reference point that does not depend on a manager's sign-off. 

That only works, however, if the values are specific enough to guide real trade-offs and if leaders visibly use them, explaining their own decisions in those terms and backing staff whose decisions were reasonably made in line with them, even when hindsight suggests a different call.

“The organisation becomes busy but less effective, and the cost of its caution is borne by the people it exists to serve.”

A loss of decision-making confidence is usually treated as a risk-management issue. I think it is just as often an institutional-memory issue. Institutional memory is what once told people what a good decision looked like: which risks really mattered, which did not, and where the boundaries actually lay. When that memory fades, the organisation retains the caution but loses the context. Rebuilding confidence depends on recovering that context, which is why institutional memory matters so much more than it first appears.

What institutional memory really means

When people hear the term institutional memory, they often think about records: policies, reports, board papers, archives and meeting minutes. These matter, but they are only part of the picture. I find it more useful to think of institutional memory as operating at three levels.

The first level is information: the formal record of what happened, held in documents, emails, policies, reports, meeting notes and databases.

The second level is knowledge, which explains why things happened. It captures the options that were considered, the risks that were debated, the assumptions that informed decisions and the lessons that emerged.

The third level, and the most valuable, is judgement. Built from experience accumulated over many years, it is the ability to recognise patterns, spot warning signs and distinguish between issues that require intervention and those that do not.

Information can be stored and knowledge can be documented, but judgement has to be developed. The distinction matters because information alone rarely answers the most important organisational questions. The critical question is often not 'What did we do?' but 'Why did we do it?', and perhaps more importantly, 'What did we learn?'

More information, less understanding

For most of history, organisations struggled because they lacked information. Today, many struggle because they have too much of it.

Think about any significant organisational decision. Behind it there may be hundreds of emails, multiple versions of advice, Teams conversations, risk assessments, board papers, meeting records and informal discussions. The information exists; the challenge is finding it, understanding it and connecting it.

Many organisations now hold more information than at any point in their history yet have less shared understanding than they realise. People spend significant time searching for documents, recreating previous analysis and reopening debates that were settled years earlier.

We have become very good at storing information, but much less effective at turning it into enduring organisational wisdom. Storage is no longer the problem. Interpretation is.

Where AI fits

AI has the potential to reshape organisational memory. Imagine being able to ask why a policy was changed four years ago, what concerns were raised when a project was approved, or how comparable cases were handled in the past. Answering questions like these once meant days of searching archives and tracking down the people who had been involved; increasingly, AI can do much of that work in minutes, connecting more documents, meetings and decisions than any individual could ever hold in their head.

Used well, this could help rebuild the very confidence this article is about. A staff member facing a difficult call could quickly see how similar matters were approached before and why, giving them context for their own judgement rather than a reason to escalate. But there is a catch: AI can only retrieve the reasoning that someone took the trouble to record. If the reasoning was never written down, no system can recover it.

There are risks too. When a summary arrives in seconds and a recommendation appears before anyone has wrestled with the problem, the temptation is to engage less deeply. The more we outsource synthesis and interpretation, the less practice people get at developing judgement, and because AI is very good at producing answers that sound right, it is easy to stop questioning assumptions and testing conclusions. AI may preserve institutional memory while quietly eroding institutional wisdom.

For an organisation that has already lost confidence, there is a further danger. The conversation can shift from 'What do we think?' to 'What did the system say?', and AI simply becomes one more place to escalate. Confidence then rests on technology rather than understanding, and the technology gets smarter while the people become less practised at exercising judgement. For regulators and other public bodies, that is not only a capability risk but an accountability one, because whoever holds the authority to decide must still own the decision and be able to explain it in their own terms. That is not a future I believe we should accept.

Remembering everything, learning nothing

An organisation can remember everything and still learn nothing, and we already see this happen. Reviews are completed, lessons identified, recommendations documented and action plans developed, and then, a few years later, similar problems emerge.

Usually this is not because the information disappeared. The reports, recommendations and records are all still available. What is missing is the process through which those lessons became embedded in the way the organisation thinks. The memory existed, but the wisdom did not.

Wisdom comes from repeated application, reflection and discussion, and from leaders who understand not only what happened but why it mattered. No AI system can fully replace that process.

Questions for boards and leaders

Boards and executive teams need to know whether people have the clarity and confidence to use organisational knowledge. Five questions matter:

1. When risk appetite changes, do people know what has changed, what has not, and where decisions should be made?

2. Are routine decisions moving upwards, and what do our decision times and escalation volumes tell us?

3. Do our values give staff enough confidence when the rules are unclear?

4. Do we record the reasoning and assumptions behind significant decisions so future leaders understand why?

5. Is AI strenghtening judgement while accountability remains with the decision-maker?

The practical response is straightforward. Define what a good decision looks like, clarify who can make it, and use organisational values when the rules are unclear. Review whether escalation is adding value, and where possible replace blanket pre-approval with checks after the event. Record the reasoning behind significant decisions, and treat AI as an aid to judgement, never a substitute for it.

The confidence to decide

For decades, organisations worried about losing information. Today, keeping information is rarely the hard part; the challenge ahead is preserving judgement.

AI may well become a powerful tool for maintaining institutional memory, but access to information is not the same as understanding. That still requires curiosity, judgement and context, and leaders who are willing to engage with complexity rather than settle for convenient answers.

The organisations that succeed over the next decade will not necessarily be those that remember the most. They will be the ones that learn the most: those that carry the lessons of experience forward, are clear about their appetite for risk when circumstances change, keep their confidence in decision-making, free their senior leaders to focus on strategy rather than escalated detail, and use AI to strengthen judgement rather than replace it.

“The greatest risk facing organisations may not be that people make the wrong decisions, or even that AI starts making decisions for us, but that, slowly and almost imperceptibly, capable people lose the confidence to make them at all. And once an organisation forgets what a good decision looks like, it does not matter how much information it can remember.”