Financial model review.

What a review is for

Someone else built the model, and a decision now rests on it. An acquisition price, a debt quantum, a board approval, a tariff submission. Review establishes whether the numbers follow from the stated assumptions, and where they do not, how much the answer moves once the errors are corrected.

Most models we review are not wrong in an obvious way. They produce plausible outputs from logic that breaks under a scenario nobody ran, or carry an assumption buried inside a formula that no reader would find by inspection. Those are the findings that change decisions.

What the review covers

Structural integrity. Whether inputs, calculations and outputs are separated, whether rows are consistent across the time series, and whether anything is hard-coded inside a calculation. We work to the FAST Standard, to which GIA is a signatory, as the reference for how a model should be built.

Logic and arithmetic. Formula consistency across every row and column, the treatment of the first and last periods, circularity and how it is resolved, and whether flags and switches behave correctly at every setting rather than only at the default.

Accounting, tax and financing. Depreciation and capital allowances, loss carry-forward, VAT and withholding treatment, working capital mechanics, debt drawdown and repayment, interest during construction, reserve accounts and cash waterfall priority.

Behaviour under stress. We run the model across the full range of its own switches and sensitivities. Models that work at base case and break at downside are common, and the break is usually in the waterfall or the debt module.

What a review does not settle

A model review tests mechanics, not commercial judgement. If the model assumes a tariff nobody will pay, the review confirms the tariff flows correctly through the calculation. Assessing whether the assumption itself is defensible is commercial due diligence, and the two are usually run together on a transaction. We say plainly which questions each answers, so that a clean review is not mistaken for a clean investment case.

Levels of review

Red flag review. A short pass identifying material structural problems and whether the model can support the decision at all. Useful before committing to a transaction or to a full review. Set out under early-stage red flag review.

Full review. Line-by-line testing against the scope above, with every finding logged.

Rolling review. Where a model is updated through a live transaction, review runs alongside each version rather than once at the end, so that errors are caught before they are built on.

Reporting

Findings are logged individually with the affected cells, the cause, the quantified effect on the key outputs, and a severity rating. Severity is set by effect on the decision rather than by how far the formula departs from good practice. A cosmetic inconsistency and a waterfall error are not the same finding, and grading them together makes a report harder to act on.

Remediation

Where the structure is sound, we remediate and re-test. Where inputs are scattered through formulas or the time series is inconsistent, remediation costs more than a rebuild, and we say so with the reasoning rather than recommending the larger piece of work by default.

Model review: common questions

How does this differ from a lender’s model audit?

A lender’s model auditor is appointed by the financing group and reports to it. Our review is commissioned by the sponsor, acquirer or authority, usually before that process starts, to find and fix what the auditor would otherwise find. The scope overlaps; the client and the timing do not.

How long does a review take?

A red flag pass on a project model is typically a few days. A full review depends on size and condition rather than on asset class, and a poorly structured model takes longer to review than a well-built one of twice the size.

Will you review a model you did not build?

That is the normal case. Where we built the model ourselves, an independent reviewer should be appointed instead, since reviewing our own work provides no assurance to a third party.

What are the most common material findings?

Debt sizing that does not resolve correctly under sensitivity, cash waterfall priority that breaks at low cover, tax treatment applied on a basis that cannot be traced to a source, and switches that produce valid-looking output at settings nobody tested.

Related expertise

• Financial modelling
• Commercial due diligence
• Early-stage red flag review
• Buy-side advisory