The prompt
Score 10 of 10, Excellent
Use it when Your spreadsheet model gives one answer, and you want to show what happens under better and worse assumptions by switching one cell.
- {{audience}}
- {{decision}}
- {{outputs}}
- {{app}}
- {{model_layout}}
- {{drivers}}
Full prompt and why it works
Prompt
Add scenarios to my spreadsheet model so that I can switch it between a base case, a best case and a worst case from one cell, and show {{audience}} every scenario's results side by side. They will use it to decide {{decision}}, so the numbers must come from the model itself, never retyped. What the model calculates, and the cells that hold the results: {{outputs}} Spreadsheet app and version: {{app}} How the model is laid out now (sheets, where the inputs are, the formulas that use them): """ {{model_layout}} """ The drivers to vary, with the base value of each, the best and worst values I have chosen, and where each value came from: """ {{drivers}} """ Build it as follows, giving the exact formulas with my sheet names and cells: 1. A Scenarios sheet with one row per driver and the columns Driver, Unit, Base, Best, Worst, Live, Source, where Live picks the value for the chosen scenario. 2. A selector cell with a dropdown (data validation) listing the scenario names, and the Live formula, for example =INDEX(C5:E5, MATCH($B$2, $C$4:$E$4, 0)). That example shows the form only. 3. The changes to the model: every formula that now uses a typed-in driver value should point to that driver's Live cell instead. List each cell to change. 4. A summary table with each result under every scenario at once. In Excel, use a one-variable Data Table (What-If Analysis) with the selector as the input cell, and put it on the same sheet as the selector, because Excel requires the input cell to be on the Data Table's sheet. In Google Sheets, which has no Data Table, give the simplest working alternative and its limits. 5. Checks, as a numbered list: with the selector on Base, every result matches the current model to the cent; and each result moves in the expected direction in the other scenarios. Keep the explanation under 400 words, outside the formulas and the layout. A good setup is one where changing a driver in one cell updates every result, and nobody has to retype a number. Do not invent driver values or ranges. Where I have not given a best or worst value, write [VALUE NEEDED] in that cell. Do not attach probabilities to the scenarios or call any of them likely. Say in one line that setting every driver to its best or worst value at once gives an extreme case rather than a realistic one, and leave that choice to me. If a driver could push a result either way, for example a higher price that brings fewer orders, ask which value belongs in the best case instead of choosing.
Variables
- {{audience}}
- Who will see the scenarios, for example "the leadership team at the budget review".
- {{decision}}
- What the scenarios inform, for example "whether to open a second location next year".
- {{outputs}}
- The results that matter and their cells, for example "annual profit in Results!C12, months of cash in Results!C15".
- {{app}}
- For example "Excel for Microsoft 365" or "Google Sheets".
- {{model_layout}}
- Sheet names, where the inputs are typed, and the main formulas that turn them into results.
- {{drivers}}
- Each driver with its unit, base value, best value, worst value and source, for example "Price per unit, USD, 40, 44, 36, last year's range".
Why it works
- What good looks like. With the selector on Base the model must match its current results exactly, so adding scenarios cannot change the numbers you already trust.
- What to avoid. Missing best or worst values become [VALUE NEEDED] and no probabilities are attached, so the scenarios say only what you chose.
- If unsure. A driver that can push a result either way, such as price against volume, comes back as a question instead of being sorted into a case by the model.