PP#6 - The Decision-Maker's Pack

You are helping me frame a decision before I make it.

The decision I need to make:

[DESCRIBE THE DECISION, WHO OWNS IT, AND THE LATEST DATE BY WHICH IT MUST BE MADE]

Context and constraints:

[DESCRIBE ANY KNOWN BUDGET, TEAM, TIMELINE, OR ORGANISATIONAL CONSTRAINTS THAT BOUND THE OPTIONS]

Using only the information I have provided, produce a structured decision frame with these five components:

1. The decision in one sentence: what is being decided, who is the decision-maker, and by when.
2. The reversibility tier: irreversible (cannot be undone), mostly reversible (cost and effort to reverse but possible), or reversible (low cost to reverse). Explain the reasoning in two sentences.
3. The criteria this decision will be judged on, ranked by importance. Maximum five criteria. Each criterion must be specific and measurable, not vague ("total year-one cost stays within GBP 50k" is a criterion; "value for money" is not).
4. The options under consideration, including "do nothing" as an explicit baseline.
5. The information currently missing that would block or significantly improve this decision. List each gap explicitly.

Output format: numbered list, plain English, no hedging language.

Do not fill in gaps by assuming context I have not provided. Do not recommend a specific option. If the decision is too vague to produce any of the five components, name the specific information missing rather than guessing.

When to use: At the start of any significant decision, before asking for analysis or recommendations. The value is in committing to specific, measurable criteria before seeing any options evaluated.

You are helping me map the stakeholders around a decision before I finalise it.

The decision:

[DESCRIBE THE DECISION IN ONE OR TWO SENTENCES]

My role in this decision:

[YOUR TITLE OR FUNCTION AND YOUR RELATIONSHIP TO THE DECISION]

Teams and functions I am already aware of:

[LIST THE GROUPS YOU ALREADY KNOW ARE INVOLVED OR AFFECTED]

Produce a stakeholder map with these four components:

1. Primary stakeholders: those directly and immediately affected, with a brief note on how.
2. Secondary stakeholders: those downstream or in adjacent functions who will bear second-order effects. Specifically identify groups that are easy to overlook in decisions of this type.
3. Veto holders and blockers: those with formal or informal power to prevent or significantly delay this decision. For each, note the source of their influence.
4. Communication sequence: in what order do these stakeholders need to be informed or consulted, and what is the specific risk of getting the sequence wrong?

Output format: four numbered sections, structured list, plain English.

Do not limit the map to the groups I have already named. Do not produce generic change management advice. Do not invent specific individual names not implied by my input.

When to use: Before announcing a decision or beginning formal consultation. Most valuable when the decision crosses multiple functions, geographies, or involves external parties such as regulators, clients, or landlords.

You are acting as a rigorous challenger for a decision I am about to make.

My preferred option:

[DESCRIBE THE OPTION YOU ARE CURRENTLY LEANING TOWARDS]

My reasoning for it:

[DESCRIBE WHY YOU PREFER THIS OPTION - the logic, evidence, or intuition behind it]

Context and stakes:

[DESCRIBE THE SITUATION, THE DECISION STAKES, AND ANY RELEVANT CONSTRAINTS]

Build the strongest possible case against my preferred option. Produce your analysis in four parts:

1. The single biggest structural flaw in my reasoning: where my logic is weakest, where I am relying on an untested assumption, or where I am confusing correlation with causation.
2. Three specific and probable failure modes: concrete ways this option is likely to underperform or fail. Label each by mechanism (e.g. "execution failure", "market reaction", "financial overrun").
3. The strongest alternative option I am underweighting: the option I am dismissing too quickly, and why it deserves more weight.
4. The 90-day signal: one specific, measurable indicator that would confirm within 90 days that this decision was a mistake.

Output format: four numbered sections, direct analytical language, plain English.

Do not begin by validating or praising my preferred option. Do not present the weakest possible objections. Do not soften the critique in the conclusion. If my reasoning has a fundamental flaw, name it directly.

Voice note: If the AI's objections feel unconvincing, press it: "Build the case more forcefully. What am I most likely to be wrong about?" The AI default is to agree. This prompt breaks that default, but only if you hold it to a high standard.

You are helping me run a pre-mortem on a decision I am about to make.

The decision:

[DESCRIBE THE DECISION AND WHAT IT INVOLVES IN PRACTICAL TERMS]

The timeframe:

[HOW LONG BEFORE WE WOULD KNOW WHETHER THIS DECISION SUCCEEDED OR FAILED - e.g. 6 months, 12 months]

Assume this decision has been made exactly as currently planned and has been an unambiguous failure by the end of the timeframe. Produce the following:

1. The failure narrative: a short paragraph describing exactly how and why this decision failed. Focus on internal execution failures and predictable external reactions. Do not use black swan events.
2. The three most likely root causes: systemic or structural reasons behind the failure. For each, state why this was predictable rather than surprising.
3. Three 30-day warning signals: specific, measurable indicators that would have told us at the 30-day mark that the decision was going wrong. Each must be concrete enough to monitor.
4. One pre-emptive action: the single most important thing to do before implementing this decision to reduce the probability of the most likely root cause.

Output format: four numbered sections, plain English, no corporate jargon.

Do not invent black swan events. Do not hedge the failure narrative. Do not produce a generic risk register.

When to use: Before you commit to implementation, not during. Most effective when the decision already feels settled. For high-stakes irreversible decisions, combine with Prompts 6 and 7 before acting. If your decision relied heavily on AI analysis, consider adding this explicit failure mode: "AI-generated analysis was used as a primary input without independent verification of its key claims" - structurally predictable, not a black swan.

You are helping me map the trade-offs in a decision where several options are under consideration.

The decision:

[DESCRIBE WHAT YOU ARE DECIDING AND WHY IT MATTERS]

The options under consideration:

[LIST EACH OPTION IN A SHORT PHRASE - include "do nothing" as a baseline]

The dimensions that matter most:

[LIST 3-5 FACTORS - e.g. speed to implementation, total cost, quality, risk level, team capacity]

For each option produce these three items:

1. What this option optimises for: the dimension or dimensions where it performs best.
2. What this option actively sacrifices: the specific thing it gives up in exchange for those gains.
3. The hidden cost of the sacrifice: the downstream or long-term consequence of the thing being given up, that may not be visible at decision time.

After analysing all options, close with one sentence stating the core tension: the fundamental trade-off that no available option avoids.

Output format: one structured block per option, then the core tension sentence.

Do not recommend a specific option. Do not produce a generic pros-and-cons list. Do not minimise sacrifices to make options seem more palatable than they are.

When to use: When discussions keep returning to the same arguments, or when the team is going in circles. The core tension sentence at the end often becomes the working decision summary.

You are helping me audit the assumptions underpinning a decision I am about to make.

The decision and my current reasoning:

[DESCRIBE THE DECISION AND WHY YOU ARE CURRENTLY LEANING TOWARDS IT - include the logic, evidence, or experience you are drawing on]

Relevant context:

[DESCRIBE THE SITUATION, MARKET CONDITIONS, TEAM CAPACITY, OR CONSTRAINTS THAT INFORM THE DECISION]

Based only on what I have provided, identify the assumptions carrying the weight of my reasoning and analyse them as follows:

1. An exhaustive list of the assumptions in my reasoning. Categorise each as: factual, behavioural, or structural.
2. For each assumption, rate: Impact if false (High / Medium / Low) and Current certainty (High / Medium / Low).
3. The load-bearing assumption: the single assumption that, if proven false, would most fundamentally change the decision. State it in one plain sentence.
4. A rapid test: a specific, low-cost action I can take within 48 hours to test whether the load-bearing assumption is actually true.

Output format: assumption list as a table (Assumption | Category | Impact if False | Current Certainty), then separate sections for the load-bearing assumption and the rapid test.

Do not list generic business risks not connected to my specific reasoning. Do not infer facts I have not stated. Mark any speculative assumption clearly as speculative.

Voice note: If the load-bearing assumption surprises you, that surprise is the signal. Run the rapid test before committing resource. Note: assumptions that AI surfaces from your stated reasoning carry the same V=0 constraint as all AI output. If the load-bearing assumption is itself derived from prior AI analysis, treat it with elevated scepticism - the rapid test replaces AI-generated confidence with human-verified evidence.

You are helping me assess how reversible a decision is, so I can calibrate how much time and analysis to invest before committing.

The decision:

[DESCRIBE THE DECISION AND WHAT COMMITTING TO IT INVOLVES IN PRACTICAL TERMS]

The main consequences if it goes wrong:

[DESCRIBE WHAT WOULD HAPPEN IF THIS DECISION PROVES TO BE THE WRONG CALL - financially, operationally, or reputationally]

Analyse this decision using the following four components:

1. Classification: irreversible (one-way door), mostly reversible (two-way door with meaningful friction), or fully reversible (low cost to change course). State the classification and explain the reasoning.
2. Reversal cost: if we made this decision today and reversed it in 90 days, what would the specific cost be in capital, time, and reputational terms?
3. Diligence recommendation: based on the classification, how much information and analysis should we demand before acting? Give a concrete recommendation.
4. Unbundling option: can this decision be broken into smaller, more reversible sub-decisions that reduce the cost of being wrong? If yes, describe how.

Output format: four numbered sections, plain English, specific language.

Do not classify a decision as more reversible than it actually is. Do not recommend acting quickly on genuinely irreversible decisions. If the reversal cost is high, state it plainly.

When to use: Run this early, ideally before the Decision Frame, to calibrate how much effort the whole decision deserves. If the decision tests as fully reversible, you may not need the rest of the pack.

You are helping me create a formal record of a decision that has been made or is about to be finalised.

Decision made:

[DESCRIBE WHAT WAS DECIDED IN ONE CLEAR SENTENCE]

Decision owner:

[THE NAME OR ROLE OF THE PERSON ACCOUNTABLE FOR THIS DECISION AND ITS IMPLEMENTATION]

Date of decision:

[THE DATE THIS DECISION WAS OR WILL BE MADE]

Options that were considered:

[LIST ALL OPTIONS EVALUATED, INCLUDING ANY "DO NOTHING" BASELINE]

Chosen option:

[WHICH OPTION WAS SELECTED]

Key criteria and evidence used:

[THE MAIN FACTORS THAT DROVE THE CHOICE AND THE INFORMATION BEHIND THEM]

Using only the information above, produce a one-page decision record with these seven sections:

1. Executive summary: the decision and the core rationale in two sentences.
2. Context and criteria: the situation that prompted the decision and the criteria it was judged against.
3. Alternatives considered and rejected: each non-selected option with a brief, specific reason for rejection. Do not omit any options listed above.
4. Reversibility and review date: how hard this decision is to reverse, and when the outcome will be formally reviewed.
5. Accountability: who owns execution, who was consulted in making the decision, and who needs to be informed.
6. AI contribution: if AI tools were used in the analysis leading to this decision, state what they were used for and what they produced. If no AI tools were used, state that explicitly. Describe a specific act, not a category - "used AI to draft a stakeholder map which was then reviewed and revised by the COO" is a record; "AI-assisted analysis" is not.
7. Independent verification: what was checked by a human, independently of AI output, before this decision was made. State specifically what was verified, by whom, and what it confirmed or changed. If no independent verification was performed, state that explicitly. This section exists because AI output cannot verify itself - the record of what a human independently confirmed is the only evidence of a sound decision process if the decision is later challenged.

Output format: seven clearly labelled sections, concise prose and bullets, neutral and documentary tone throughout.

Do not write in a persuasive or advocacy tone. Do not add information not provided in the inputs above. Do not omit alternatives. Do not use governance language such as "human in the loop" or "AI-assisted with oversight" in sections 6 or 7 - describe the specific act, not the category.

When to use: Within 24 hours of the decision being made, while the reasoning is fresh. Sections 6 and 7 are the governance additions that make this record defensible rather than merely descriptive. The standard is not whether you had a process - it is whether you can show, from existing records, what AI contributed and what a human independently confirmed before you proceeded.