Hiring a CFO to review a financial model: agree what they will test

Send prospective reviewers a description of the model, its owner and the decision it supports. Identify the exact file or system version, the main outputs and the deadline. A formula review of one workbook differs from reviewing the commercial assumptions, source data and linked systems that feed the forecast.

Ask the reviewer to state what independence means for this assignment. Someone who built the model has different prior involvement from a reviewer approaching it for the first time. ICAEW's spreadsheet principles recommend review and testing proportionate to risk and recognize the limits of self-review. Agree the arrangement that fits the decision rather than relying on the word independent in a proposal.

List the areas that could change management's decision: debt repayment logic, tax assumptions, working capital, consolidation, timing or a particular valuation output. Ask which procedures will examine each area and which specialist judgments remain outside the engagement. A reviewer may test that an approved tax assumption is applied consistently without being engaged to determine whether that assumption is legally correct.

Require a distinction between formula checks, data reconciliation, scenario behavior and assumption challenge. A workbook can calculate exactly as designed while using an unsupported growth rate. Conversely, a reasonable commercial plan can be misrepresented by a broken formula. The proposal should make clear which of those problems the reviewer is being paid to investigate.

Give the reviewer a stable copy with the supporting inputs and necessary access. Record the review version and keep a change log for updates made while testing proceeds. If the operating team continues revising the model, agree how the reviewer will identify and assess the changed parts.

The engagement should explain how external links, hidden sheets, macros or licensed add-ins will be handled where present. Do not assume a review of a PDF output covers the underlying workbook. Identify any component the reviewer cannot access and ask how that limitation affects the usefulness of the findings.

Each finding should identify the affected location, observed issue, potential consequence and proposed owner. Ask for a prioritization based on the decision, rather than a long list of cosmetic preferences. The team needs to know which problems could materially change the answer and which improve maintainability.

Agree whether corrections are made by the original model owner, the reviewer or another specialist. Include retesting of significant changes and a record of unresolved findings. For a hypothetical model, correcting a debt-interest formula could also change cash, tax and debt balances; the review should follow those consequences rather than confirm only that one cell now looks different.

Ask what form the final report will take, who may use it and what limitations will be stated. A commercial model review should not be described as a statutory audit or a guarantee of future results unless an appropriately qualified provider has separately agreed the relevant professional service and reporting requirements.

Accept the work when the reviewed version, procedures, findings, corrections and remaining limitations are clear. Retain the final model and report together. Later changes can require further review, so specify the circumstances that trigger a new assignment. The useful outcome is a better-understood model and a traceable review, not an assurance label that conceals what was actually tested.