Financial model audit

Roughly nine in ten substantial spreadsheets contain at least one material error. That statistic has been stable across academic and industry studies for two decades, and nothing about the last few years has improved it. The question is never whether your model contains errors. It is whether you find them, or a diligence analyst does.

The difference matters more than the error itself. A mistake you disclose and correct is a housekeeping item. The same mistake found by an investor's analyst in week three of diligence becomes a question about everything else in the workbook — and re-opened diligence is how rounds slip a quarter.

What the review covers

Mechanical integrity. Every calculation block is read cell by cell. Inconsistent formulas across a row, ranges that stop one column short, SUMs that double-count a subtotal, IFERROR wrappers hiding a genuine failure, circular references resolved by iterative calculation, volatile functions, and hardcoded constants sitting inside formulas where nobody will ever find them.

Structural integrity. Whether the three statements genuinely articulate, whether the balance sheet ties without a plug, whether cash flow reconciles to the movement in the cash balance, whether sign conventions are consistent, and whether the model still behaves when a driver is pushed to a plausible extreme rather than a comfortable one.

Logic and assumptions. Whether the revenue build actually produces the revenue claimed, whether churn and expansion are applied to the right base, whether headcount costs include the loading that real employment carries, whether working capital moves with volume, and whether the growth, margin and retention assumptions sit inside the range a sector benchmark would support. This is where the expensive problems usually live — a mechanically flawless model can still be wrong.

What you receive

An error register listing every issue with its location, a plain description, the quantified impact on the headline outputs, a severity rating and a recommended fix. Issues are ranked by materiality, so the two things that matter are at the top rather than buried among thirty formatting notes.

Alongside it: a structural assessment saying plainly whether the model should be repaired or rebuilt, and a summary letter written for sharing — scope, method, findings and resolution status — that sponsors put in the data room next to the model itself.

Timing and fee

Reviews run five to ten business days and start at $2,500 for a typical operating or development model. Fee scales with tab count, complexity and whether corrections are applied. An expedited three-day review is available at a premium when a data room is already open.

Where the review concludes that the structure cannot be economically repaired, the audit fee is credited against a model build. The point of the engagement is a model you can defend, not a longer list of problems.