Sensitivity Tables (One- and Two-Variable)

When to use

Use when a strategy or finance model has a single key output and you want to show how that output responds as one or two drivers vary across a range. Typical drivers are price, volume, discount rate, growth, churn, or cost per unit. This skill BUILDS a downloadable .xlsx from scratch in the Claude app (openpyxl plus numpy), with live formulas and a recalc-and-verify pass. It is not the Claude for Excel add-in.

What it builds

A single Sensitivity tab that attaches to an existing model. It references the model's input cells and output cell, lays out one-variable and two-variable grids, and color-scales the result block. If the source model lives on another sheet, the grid links to it with cross-sheet references so it stays live.

Build workflow

1. Identify the output cell (for example Model!B40 holding NPV) and the one or two input cells that drive it (for example Model!B6 discount rate, Model!B7 growth).
2. Confirm the output is a live formula chain back to those inputs. If the output is hardcoded, stop and wire the model first; a Data Table over a constant returns the constant.
3. Choose input ranges around the base case (for example base discount rate 10 percent, sweep 6 to 14 percent in steps of 2).
4. Build the native Excel Data Table structure: corner formula, input axes, and the {=TABLE(r,c)} array (see spec).
5. ALSO build the code-computed grid as a robust fallback: re-evaluate the model in Python across every input combination and write the resulting values as a labeled matrix.
6. Apply a 3-color scale to the result block and label both axes with the input cell names.
7. Add the Checks block.
8. Recalculate the workbook in a real engine (headless LibreOffice).
9. Verify zero formula errors (#REF!, #DIV/0!, #VALUE!, #NAME?), confirm the checks pass, then deliver the .xlsx.