Cost-Benefit Analysis Checklist

State the decision in one sentence: buy vs. lease the warehouse, replace the ERP, hire two AP clerks vs. license Bill.com. A vague question ("should we improve AP?") produces a vague analysis. The CFO or requesting department head signs off on the wording before any modeling begins.

Pick a horizon that matches the asset's useful life or contract term — typically 3, 5, or 7 years. Use the company's WACC as the discount rate, or the hurdle rate set by the board. Document both assumptions in the model so reviewers don't have to reverse-engineer them.

Identify who owns the operational outcome (sponsor) and who has signing authority per the capex approval matrix. Projects above the CFO's threshold typically route to the CEO or board; getting this wrong forces a re-presentation.

Capture headcount caps, debt covenants, available cash, lease vs. buy preferences, and any timing constraints (fiscal-year close, lease expiry). Constraints that surface late in the analysis force rework.

Capture year-by-year hardware, software licenses, implementation fees, training, materials, and direct labor. Tie each line to a vendor quote, SOW, or signed offer. "Estimated" lines without a source are the first thing reviewers challenge.

Allocate IT support, facilities, management time, and integration work. Use the company's standard overhead rate if one exists; otherwise document the allocation method. Don't bury these in a single "contingency" line.

Data migration, parallel running, productivity dip during cutover, and contract termination fees on the legacy system. Common omission: the 6-12 weeks of reduced output while staff learn the new tool.

Model new revenue, FTE reduction, vendor consolidation savings, and reduced error rates with dollar values. Tie each benefit to a measurable KPI — DSO improvement, close-cycle days reduced, AP invoices per FTE — so post-implementation tracking is possible.

Faster month-end close, fewer audit adjustments, improved forecast accuracy, lower turnover. Convert hours saved to dollars at fully-loaded rates; mark these benefits clearly as "soft" so the board sees what's hard money vs. estimated.

SOC 2 readiness, audit-fee reduction, segregation-of-duties improvements, scalability for the next funding round. Don't quantify what you can't defend; describe these qualitatively in the recommendation memo.

The controller, AP manager, or department head whose team will deliver the savings signs off on the numbers. Finance-only benefit estimates that operations hasn't endorsed get challenged in committee and lose credibility fast.

Lay out cash inflows and outflows by year over the analysis horizon. Apply the discount rate from the scope step. Keep assumptions on a separate tab so reviewers can flex them without breaking formulas.

Report all three: NPV at the company's WACC, IRR for comparison vs. hurdle rate, and undiscounted payback. The board usually anchors on payback for risk and NPV for value creation.

Flex the top three assumptions ±20% — typically adoption rate, FTE savings, and implementation cost. Build a tornado chart so the partner sees which assumption breaks the case if it's wrong.

If NPV is negative or marginal at the hurdle rate, the next phase forces a serious look at alternatives before recommending. A positive NPV with a strong sensitivity profile lets the team move toward recommendation.

Always include the do-nothing case and at least one lower-cost alternative. For software decisions: best-in-class vs. suite vs. status-quo-with-process-fix. Boards rarely approve a proposal without seeing what was rejected.

Quantify the cost of inaction — continued manual work, audit findings, growing technical debt, lost revenue. The do-nothing case is often understated; it's what makes the proposed option look good or bad.

Use a weighted matrix covering NPV, payback, implementation risk, vendor stability, and strategic fit. Document the weights before scoring so the comparison can't be reverse-engineered to favor a preferred option.

One to two pages: decision question, recommendation, NPV / IRR / payback, top three sensitivities, alternatives considered, implementation timeline, and approval requested. Save the supporting model as an attachment, not as the primary deliverable.

Walk through the model assumptions, the sensitivity tornado, and the alternatives matrix. Capture pushback as model revisions before the formal approval meeting — never the day of.

CFO, CEO, or board per the capex approval matrix. Bring the model open in case someone wants to flex an assumption live. Document the decision and any conditions in the meeting minutes.

Pick the two or three KPIs that prove the benefits landed — close-cycle days, AP cost per invoice, DSO. Schedule the 6-month and 12-month look-back so the firm builds the discipline of comparing forecast vs. actual on capex.

If NPV is negative at the hurdle rate, document the analysis, archive the model, and notify the sponsor with the reasoning. A clean kill protects credibility for the next request — and prevents the same proposal coming back unchanged in six months.