Most decisions cannot be settled
A course on decision-making that emphasised evidence would misrepresent what senior roles in this sector involve.
The consequential decisions are frequently about the future under genuine uncertainty. Whether a regulatory regime will tighten. Whether a market's competitive intensity will moderate. Whether an acquired team will stay. Whether a supplier will continue investing. Whether a product will find an audience. Whether a person will grow into a role.
No amount of analysis converts these into facts. Treating them as though it could produces delay, which is itself a cost, and false confidence, which is worse.
What good practice provides is not certainty but better-structured judgement, and that is a lower and considerably more achievable standard.
Reversibility as the organising principle
The framework introduced in the Operations Strategy course, restated here because it is the most useful single discipline available to a leader.
Reversible decisions should be made quickly and corrected if wrong. Promotional structures, page layouts, process changes, most hiring at junior levels, and anything that can be tested. Extensive deliberation over these wastes the scarcest resource in the organisation.
Irreversible decisions warrant the analysis, because there is no opportunity to learn from doing. Platform selection, market entry with licensing commitments, acquisitions, senior appointments, brand positioning and long-term contractual obligations.
Most organisations invert this. They deliberate over campaigns that could have been tested in a fortnight and commit rapidly to platforms they will live with for a decade, usually because the second decision came with commercial urgency and the first came with an approval process.
The practical intervention is to ask, of any decision reaching a leader, how easily it could be undone, and to route it accordingly. Decisions that are cheap to reverse should be delegated with clear boundaries rather than escalated.
What to do when evidence will not settle it
For the decisions that require judgement, several practices improve the quality without pretending to resolve the uncertainty.
Make the assumptions explicit. A decision resting on a market producing a given revenue at a given acquisition cost should state that. Assumptions unstated cannot be checked, cannot be challenged and cannot be reviewed when they fail.
State the range rather than the point. A forecast presented as a single number conveys precision that does not exist. A range with the reasoning behind its bounds is more honest and more useful.
Identify what would change the answer. Naming the conditions under which the decision would be wrong makes it reviewable and frequently reveals that those conditions are more likely than the proposal assumed.
Seek the strongest opposing case. Not a token objection but a genuine attempt to argue the other side, ideally by someone who believes it. Proposals reaching a leadership team have usually passed through filters that removed the objections.
Check the base rate. How often do market entries of this kind succeed? How often do acquisitions deliver their case? The general frequency is a useful check on an estimate derived entirely from the specific circumstances, which invariably feel exceptional.
Consider the downside independently. Not as a probability-weighted contribution to an expected value but as a question about whether the organisation could absorb it. In this sector, decisions with acceptable expected value and unacceptable worst cases are specifically dangerous, because the worst case frequently involves the licence.
Decide. Analysis has diminishing returns and delay has costs. A decision made on adequate reasoning at the right time beats a better decision made late.
Decision records
The single most valuable practice available and among the least used.
A decision record states what was decided, why, on what evidence, what was uncertain, what was assumed, who decided, and what would prompt a review.
Its value appears later. Six months or two years after a market entry, a platform selection or a structural change, the ability to see what was actually believed at the time converts the outcome into information.
Without it, the organisation relies on recollection, and recollection is unreliable in a specific direction: it reconstructs prior reasoning to match what happened. Successes are remembered as having been foreseen. Failures are remembered as having been driven by circumstances nobody could have predicted. Both prevent learning.
The record also has practical value for individuals holding personal regulatory accountability, as the first lesson of this course noted. A decision taken under pressure and undocumented is one defended from memory against a written record assembled by someone else.
The objection is that recording decisions takes time and creates a document that could be used against people. The first is true and small. The second reveals something about the organisation, since a culture where a record of honest reasoning is dangerous has a problem larger than its documentation practice.
Biases that recur here
Naming these in advance is more effective than attempting to reason around them in the moment.
Escalation of commitment. Continuing to invest because of what has already been committed. Appears as markets not exited, platforms not replaced, programmes not cancelled and acquisitions not written down. The correction is to ask whether you would start this today.
Confirmation seeking. Gathering evidence supporting a preferred conclusion rather than evidence that would discriminate. Appears in analysis commissioned to justify decisions already made, which is detectable by asking whether a contrary finding would change anything.
Optimism in forecasting. Systematic across industries and pronounced in market entry cases. The correction is base rates and explicit downside scenarios.
Availability. Overweighting recent or vivid events. A competitor's visible success in a market makes entry seem more attractive than the general record supports.
Survivorship. Learning from what worked without examining what did not, which produces conclusions that do not discriminate. Common in creative and commercial decisions alike.
Deference to seniority. The most senior view prevailing regardless of who has better information. Particularly damaging in this sector where specialist judgement in trading, compliance and mathematics is genuinely superior to generalist judgement.
Groupthink in homogeneous teams. Addressed in the second lesson of this course.
Short-horizon preference. The structural bias described in the first lesson, where the person under quarterly pressure will not experience the multi-year consequence.
Deciding well as a group
Most consequential decisions in an operator are made by groups, and group process determines whether the individual disciplines above survive.
Separate divergence from convergence. Generate options and criticism first, decide second. Groups that begin by evaluating the first proposal generate fewer alternatives.
Have people commit before discussing. Written positions circulated in advance reduce anchoring on whoever speaks first and prevent quiet agreement with the senior view.
Assign someone to argue against. Formally, so that dissent is a role rather than a personal position.
Let junior people speak first. Once a senior view is stated, the range of expressed opinion narrows sharply.
Distinguish the decision from the discussion. Groups can discuss usefully and decide badly if it is unclear who is deciding and when.
Record the dissent. Where the group was divided, the record should say so, because a decision recorded as unanimous when it was not misrepresents what the organisation knew.
Learning
The final point, and the one that matters most over time.
The purpose of everything in this lesson is not to be right more often about individual decisions. Under genuine uncertainty, a well-reasoned decision can turn out badly and a poor one can turn out well, and over any short horizon outcomes tell you little about decision quality.
The purpose is to build an organisation that notices when it was wrong and adjusts. That requires decisions recorded, assumptions stated, reviews scheduled, and a culture where being wrong is survivable enough that people acknowledge it.
An operator with that capability improves. One without it repeats, because it cannot establish what it previously believed and therefore cannot identify what it got wrong. Across a decade, the difference between those two organisations is substantial, and it is entirely a consequence of practices that cost very little and are consistently deprioritised.
Speed and its value
A counterweight, since the disciplines above could be read as an argument for slower decisions.
Delay has costs that rarely appear in any analysis. Opportunities close. Competitors move. Teams waiting on a decision cannot proceed. Uncertainty is corrosive to morale. And a decision made late may be worse than a slightly inferior decision made in time, because circumstances have moved.
The resolution is the reversibility principle rather than a general preference. Move quickly on what can be undone. Take care with what cannot. And recognise that most decisions in an organisation fall into the first category while most of its deliberation is spent there.
A practical test for a leader is to look at what actually consumes leadership meeting time. If it is dominated by decisions that could be delegated with clear boundaries and corrected if wrong, the organisation has a routing problem rather than a decision-making problem, and fixing it releases capacity for the decisions that genuinely require senior judgement.
The related discipline is deciding when to decide. Some decisions benefit from waiting because information will arrive. Others do not, and waiting is avoidance dressed as prudence. Naming which of the two applies, explicitly, prevents a good deal of drift.
Communicating a decision
Making a decision and having it take effect are different things, and the second depends on how it is conveyed.
State the decision plainly. Ambiguity about what was decided produces different interpretations across the organisation.
Give the reasoning. People implementing a decision encounter situations it did not anticipate, and they can only extrapolate if they understand the intent.
Acknowledge the alternative. Where a genuine option was rejected, saying why respects the people who argued for it and prevents the decision being relitigated.
Name what is uncertain. A decision presented with false confidence loses credibility when it encounters difficulty.
Say what would change it. Which converts a decision into something reviewable rather than something defended.
Follow up. Reporting what actually happened, including where the decision turned out badly, is what makes the next one credible and is the mechanism by which the organisation learns.
Leaders who communicate decisions well find that fewer of them require enforcement, because people who understand the reasoning implement the intent rather than the letter.
Decisions specific to this sector
Several recurring decision types in gambling have characteristic difficulties worth anticipating.
Market entry. Irreversible in practice once licensing commitments are made, decided on information that is systematically optimistic, and subject to a regulatory trajectory nobody can predict. The disciplines that help are base rates on entry success, fully loaded economics, explicit trajectory scenarios and a pre-committed exit criterion.
Platform selection or migration. The least reversible decision most operators make, frequently taken under commercial urgency, and lived with for a decade. The discipline is to weight supplier viability and roadmap alignment above current feature comparison, since features can be added and a supplier that stops investing cannot be fixed.
Whether to serve a grey market. A decision with commercial upside and option value cost elsewhere, as the Operations Strategy course set out. It should be taken explicitly at board level rather than accumulating through the absence of a decision.
Whether to restrict or close a valuable customer. The decision where commercial and protective considerations conflict most directly, and where the structures described in the incentives lesson determine whether it is made well. It should not sit with anyone whose earnings depend on the answer.
Whether to launch a feature that performs and is questionable. The product judgement described in the Product Innovation and Game Studio courses. The useful test is whether the design would be defended publicly.
Whether to self-report a failure. Almost always yes, and the decision is difficult in the moment because the consequences of reporting are immediate and the consequences of not reporting are contingent. The base rate strongly favours reporting.
Whether to slow down. Covered in the growth lesson, and characteristically unpopular at the time and obvious afterwards.
Each of these has a shape that recurs, and leaders who recognise the shape make the decision better than those encountering it as novel.
Judgement is not a lesser mode
A closing point of framing.
Discussion of decision-making frequently implies a hierarchy in which evidence is the proper basis and judgement is what remains when evidence is unavailable. That framing misdescribes senior work in this sector.
The decisions that matter most are about the future under genuine uncertainty, and they will be made on judgement regardless of how much analysis precedes them. Pretending otherwise produces analysis that dresses a judgement in numbers, which is worse than the judgement alone because it obscures what is actually being decided.
What distinguishes good judgement from poor is not the presence of data. It is whether the assumptions were made explicit, whether the opposing case was genuinely considered, whether the downside was assessed on its own terms, whether the base rate was checked, whether the decision was recorded, and whether it will be revisited when the assumptions fail.
Those practices are available for any decision, evidenced or not, and they are what this lesson has been about.
Building the practice
To make this operational rather than aspirational, what a leadership team can actually adopt.
A one-page decision record for significant decisions, covering what was decided, why, the key assumptions, what was uncertain, who decided and what would prompt review. Kept short enough that it happens.
A reversibility test applied at the point a decision reaches the leadership team, routing the reversible back down with clear boundaries.
A standing challenge role in significant decisions, rotated, so that arguing the opposing case is a function rather than a personality.
Pre-reading with written positions, so that discussion begins from independent views rather than converging on whoever speaks first.
A scheduled review of past decisions against their recorded assumptions, perhaps twice a year, examining a handful rather than all of them.
Base rate collection, meaning the organisation records how its own market entries, acquisitions, product launches and hires have actually turned out, so that future estimates have something to anchor against.
None of this requires new systems or additional people. It requires the discipline to do it when the immediate pressure is to move on, which is the same constraint that applies to everything else in this course.
The return appears over years rather than months. An organisation with three years of decision records and a habit of reviewing them knows things about itself that no amount of analysis can substitute for, and it makes measurably better decisions as a result.