Seven of them, and the useful ones share a property: they ask for a disagreement rather than a document. A prompt that asks for a plan returns something agreeable that commits to nothing.
Most prompt lists for this are a way of getting a plan written faster. A plan written faster is not the problem any leadership team actually has.
The seven
1. The case against.
Here is our plan for next quarter. Make the strongest possible argument that it will fail. Do not soften it and do not offer a balanced view. Assume you are the person who will be proved right.
This is the highest-return prompt on the list. It reliably surfaces two things everyone in the room was aware of and nobody wanted to say first.
2. What would have to be true.
For this objective to be achieved, what would have to be true about our customers, our team and our market? List each assumption separately and mark which ones we have evidence for.
Turns a confident plan into a list of testable bets. The ones with no evidence beside them are the plan's real risk.
3. The implicit refusal.
Given this list of priorities, what are we implicitly choosing not to do? Name the things that will quietly get no attention.
Strategy is refusals. This prompt makes them visible without anybody having to propose cutting something.
4. Sharpen one objective.
Rewrite this objective so it has a measure that can go down as well as up. If the measure can only increase, say so and suggest a different one.
The single most common defect in a plan is an activity measure dressed as an outcome.
5. The stall prediction.
Of these commitments, which three are most likely to be unchanged in six weeks, and why? Pay particular attention to anything needing cooperation between teams.
Cross-team commitments are where execution breaks, which Sull, Homkes and Sull established in Harvard Business Review in 2015. A model is reasonable at spotting which ones they are from the wording alone.
6. The comprehension test.
Read this plan as a team leader who was not at the offsite. What would you think your team is supposed to do differently on Monday?
If the answer is vague, the plan is vague, and you have found that out before everyone else does.
7. The update draft.
Here is what changed on these commitments this fortnight. Write a short progress note for the leadership team. State what moved, what did not, and what needs a decision. No encouragement.
Tedious, necessary, and the first thing to decay when people get busy.
What makes the difference
| Weak prompt | Why it fails |
|---|---|
| "Write a strategic plan for a B2B SaaS company" | No constraints, so you get the average of everything ever written |
| "What are good OKRs for a sales team?" | Produces twelve plausible ones and no basis to choose four |
| "Is our strategy good?" | It will say yes, thoughtfully |
| "What is the market size for X?" | A confident number with no source you can open |
The fix in every case is context. Paste your real numbers, real customer language, what you tried last year and what actually happened. The quality of a planning prompt is almost entirely the quality of what sits underneath it.
Where the prompts stop being enough
All seven produce an artefact in a chat window. Six weeks later the plan has moved, the window has not, and nobody has the thread anymore.
That is the gap between planning and execution, and it is not a prompting problem. It needs the plan to live somewhere every owner can update, with one name against each commitment and a rhythm short enough to catch a stall. Empiraa GPS is built for that, with ANI inside the platform so the drafting and summarising sit on top of a record that is actually current rather than on a transcript of a conversation from six weeks ago.
Related reading: the best AI for strategic planning, and the part it cannot do, and strategy execution best practices that survive a quarter.
See how to write key results, and what an AI assistant can see in your strategy.
