Financial model review

Hardcoded numbers in financial model formulas

Review hardcoded numbers inside Excel financial model formulas. Use a worked example to distinguish changing assumptions from intentional numeric constants.

An adjustable dial inside a spreadsheet beside a separate input tile

A hardcoded number inside a formula is a literal value entered directly into that calculation. It may be an intentional conversion or an assumption that should be visible elsewhere. Review its meaning and maintenance needs before replacing it: the presence of a number does not make a formula wrong.

A worked example: one rate, several places

Suppose the invented cell Forecast!D18 contains:

=C18*0.21

Here 0.21 is a fixed multiplier. It could represent a particular charge, cost share or other rate; this example does not prescribe any real tax or financial assumption. Microsoft distinguishes directly entered constants from references in its formula overview.

Ask three questions: what does this rate mean, who approved it, and where else is it used? If it is repeated across periods, editing one formula could leave the rest unchanged. A matching literal elsewhere is only a lead for review: two identical numbers can serve different purposes.

One possible redesign is to put the intended rate in Inputs!B6, label the adjacent cell clearly, and reference it:

=C18*Inputs!$B$6

This concentrates maintenance in one input. The absolute reference anchors the input when copying the formula. It does not verify that the rate is correct or that every calculation should use the same assumption.

Choose which numbers need their own inputs

Prioritize changing business assumptions: scenario rates, discounts, prices and recurring cost factors. For each one, record its description, units, source and intended period. Test the redesigned formula in Excel against the original before making wider changes.

A divisor such as =C18/12 might intentionally convert an annual amount to equal monthly amounts. Its appropriateness depends on the model; it should not be automatically replaced. Numbers can also serve as function controls: the second argument in =ROUND(C18,2) sets decimal precision rather than an economic assumption.

Named cells can improve readability, and Excel can define names for ranges or constants. Microsoft explains how names work. Use names thoughtfully: a readable name alone does not establish correct scope, a correct reference or scanner support.

How to read a coefficient advisory

When a report includes an embedded-coefficient Finding, inspect the saved formula, literal and exact sheet/cell. Treat the Finding as a review prompt. An advisory about a fixed coefficient does not reduce the provisional saved-evidence score by itself.

The check is deliberately narrower than finding every number: supported multiplicative factors and divisors are considered, while neutral values, ordinary function-argument numbers and unsupported formula structures have different treatment. A skipped formula is a limitation, not a pass.

FinModelCheck does not infer the business purpose of a coefficient or recalculate the model. The example report shows how evidence is presented. After using the checker, download the free detailed PDF and compare the reported sheet, cell and formula with your workbook in Excel. See also external workbook links and saved Excel errors.

Short answers

Should every number be moved to an assumptions sheet?

No. Start with meaningful assumptions that change or need explanation. Mechanical constants can remain inline when their purpose is clear.

Does replacing a literal with a cell reference fix the model?

It can improve maintainability. You still need to verify the input, units, formula logic and affected periods.

Does no advisory mean the model has no hardcodes?

No. A narrow check of supported multiplicative factors and divisors cannot establish the absence of all constants across every Excel feature. Review other formulas and features separately.

Keep reviewing

Continue with a related guide.

Browse all guides