Editorial guidelines & model review process

1. Who writes the models

Models are written by analysts who have underwritten the asset class they are modelling. Real estate models are authored and maintained by our Head of Real Estate Models; e-commerce models by our Head of E-commerce Models. Neither writes outside their domain, and every model carries the name and role of the person responsible.

2. Independent review before publication

No model is published without a second-person review by our Financial Model Auditor, whose role is specifically to disagree. The reviewer tests four things:

• Formula correctness: the JavaScript calculation and the Excel formula must produce the same result, cell for cell.
• Sign convention integrity: outflows negative and inflows positive throughout, with no accidental double-negatives in the waterfall.
• Edge cases: zero interest rate, 100% down payment, zero revenue, zero lead time, blank inputs, and division-by-zero paths.
• Cross-tab linkage: every cross-sheet reference resolves to the intended cell after the workbook is regenerated with different inputs.

3. Formula publication

Every formula used on a page is displayed in plain text on that page, in a monospaced panel with the variable names spelled out. There are no proprietary adjustments, hidden multipliers or unexplained constants. If a model applies a simplifying assumption — for example, a 22% annual inventory holding cost, or a 5% shrink allowance on returns — the assumption is labelled and editable.

4. Sources and defaults

• Fee schedules come from the platform’s own published rate card and are dated in the model.
• Benchmark ranges (cap rates, margins, service levels) are compiled from published industry surveys and attributed on the page that uses them. Every default, its basis and its source are listed on the data sources page.
• Default values are illustrative examples for a mid-market scenario, never optimal or recommended inputs.
• Where a figure varies significantly by market or category, we say so in the page text rather than hiding it in a footnote.

5. Update cadence

• Marketplace and payment fees: reviewed quarterly and after any announced change.
• Tax-related structure (Schedule E mapping): reviewed annually with a CPA.
• Real estate benchmarks: reviewed twice yearly, or when rate environments move more than 100 basis points.
• Every page: the workbook is regenerated and re-verified after any model change, and the “Updated” date changes accordingly.

6. Corrections policy

Report an error to [email protected] with the page URL, the inputs you used and the output you expected. Verified calculation errors are fixed and published within two business days. Substantive corrections are noted on the page with the date. We do not silently change numbers.

7. What we will not publish

• Any output presented as a prediction of a specific asset’s future value.
• Tax advice, or a model that claims to determine tax liability rather than organise categories.
• Paid placements disguised as recommendations, or rankings influenced by affiliate economics.
• Models we cannot explain line by line on the page.

8. Accessibility and performance commitments

Calculators are keyboard navigable, label every input, expose results as text rather than colour alone, and reserve fixed space for advertising so content never shifts while you read. Core Web Vitals are treated as an editorial standard, not an afterthought — a model that is slow to load is a model that does not get used.

9. Who reviews the models

SheetModel does not publish invented author credentials. Named review bylines appear on a model only when a real, identifiable person has reviewed it and agreed to be credited, and they are added to the organisation’s structured data at the same time. Until then, each model stands on the sourcing discipline above: published formulas, attributed defaults, and a source registry that anyone can audit.

Questions about the process? See the about page or contact us.