How to Document a Financial Model

Documentation is what lets someone else, or you in six months, pick up a model without reverse-engineering it. Good documentation is partly written and partly visual: an assumptions tab and notes, plus color that shows at a glance which cells are inputs and which are calculations.

Centralize assumptions on one tab

Pull every driver into a single assumptions tab with clear labels and units. Formulas elsewhere reference these cells, so there is one place to read or change any input.

This is the single highest-value documentation step. A reader who finds all assumptions in one labeled list understands the model's levers before reading a single formula.

Color-code cells by role

A long-standing modeling convention uses color to show what a cell is. Inputs are blue, formulas are black, links to other sheets are green, and links to other workbooks are a separate color. The reader learns the legend once and reads the whole model faster.

Doing this by hand means selecting each input and recoloring it, which is tedious and drifts out of date the moment someone adds a row.

Apply the color convention automatically

ModelMint's Color Coder paints cells by role in one pass: constants in blue, formulas in black, cross-sheet links in green, externals in purple, and errors in red.

Because it reads what each cell actually is, the coloring stays accurate, and you can re-run it after edits to keep the documentation current. A reviewer can tell an assumption from a calculation without clicking into a single cell.

1. Select the range or the whole sheet you want to document.
2. Run Color Coder to paint each cell by its role.
3. Add a small legend on the sheet so first-time readers learn the colors.
4. Re-run after structural edits so the colors stay accurate.

Add notes where the logic is non-obvious

Use cell comments or notes sparingly, only where a formula's intent is not obvious from its structure. Document the why, not the what; a reader can read the formula, but cannot read your reasoning.

Good candidates are unusual conventions, deliberate circular references, and any assumption with a source you want to cite. Avoid noting the obvious, which just adds clutter.

Write a short readme tab

A single readme or cover tab that states the model's purpose, the as-of date, the key outputs, and any caveats saves the next reader an hour. Keep it to a screen.

List where assumptions live, what the color legend means, and any known limitations. This is the map that turns a spreadsheet into a model someone else can trust.