DCF Model Audit Checklist: 30 Tests for Valuation

DCF Model Audit Checklist: 30 Tests for Forecasts, WACC and Terminal Value

A discounted cash-flow model should not be trusted because the spreadsheet calculates without errors. Trust comes from six audit gates: source integrity, accounting-to-cash conversion, forecast logic, discount-rate consistency, terminal-value discipline and decision controls. Each gate contains five tests, creating a DCF-30 audit that a reviewer can apply before using a valuation in an investment, acquisition, capital-allocation or portfolio decision.

This guide does not teach DCF from the beginning. It assumes a model already exists and answers the question learners repeatedly ask after building one: how do I find the defects that matter? It includes a scoring rule, a defect log and a worked review. The objective is not to force one “correct” value. It is to determine whether the model is traceable, internally consistent, economically plausible and honest about uncertainty.

What a DCF audit can and cannot establish

The general DCF idea is straightforward: estimate future cash flows and discount them at a rate appropriate to their risk. NYU Stern's Aswath Damodaran describes the core relation as expected cash flow divided by a risk-adjusted discount process over time. The hard part is that forecasts, risk and long-run economics are uncertain.

An audit can establish whether:

historical inputs trace to authoritative statements;
the model converts accounting results to the correct cash-flow definition;
operating assumptions are connected rather than independently convenient;
cash flows and discount rates use consistent claimholder, inflation, currency and tax bases;
terminal value follows a defensible steady state;
sensitivities expose—not conceal—the assumptions driving the answer;
the decision memo represents the range and limitations fairly.

An audit cannot prove that a forecast will occur, certify that a security is attractive or remove judgment from valuation. A model may pass every mechanical check and still be wrong because the business narrative fails.

Prepare an audit copy before testing

Do not review the only working file. Freeze a read-only copy and record:

model name, version and hash or controlled filename;
valuation date and currency;
author and independent reviewer;
cash-flow method: FCFF, FCFE, dividend or another explicit approach;
source documents and their dates;
last actual period and first forecast period;
base, downside and upside scenario definitions;
decision that the model will support.

Create a defect log with these fields:

ID	Gate	Test	Severity	Evidence	Effect on value	Owner	Resolution	Status
D-01	Discount rate	Nominal/real consistency	Critical	Forecast excludes inflation; WACC is nominal	Directionally over-discounts	Model owner	Restate forecast or rate consistently	Open

Use severity deliberately:

Critical: model basis is inconsistent, valuation cannot be relied on.
High: likely material effect or unsupported key assumption.
Medium: decision usefulness is reduced but the model can be interpreted with correction.
Low: presentation, documentation or minor control weakness.

Do not close a defect because the value “looks reasonable.” Close it when evidence or a controlled correction resolves the cause.

Gate 1: source integrity

Test 1 — historical-statement trace

Trace revenue, operating profit, tax, depreciation, capital expenditure, working capital, debt, cash and shares to the relevant public filing or authorized internal ledger. The SEC explains that the balance sheet is a point-in-time view, while income and cash-flow statements cover a period. Confirm that the model has not mixed a quarter-end balance with a full-year flow without a deliberate adjustment.

Pass: every material historical input has source, period, unit and transformation note.

Test 2 — unit and scale consistency

Check currency, thousands versus millions, percentages versus decimals and shares versus fully diluted shares. A model can be structurally correct and still fail by a factor of 1,000.

Pass: units appear in headers and automated checks compare key totals to source values.

Test 3 — period and calendar alignment

Determine whether fiscal years align across the company, peers and market inputs. If a company reports in June and peers in December, comparisons and forecast periods may need calendarization. Identify stub periods and ensure discount timing follows actual dates.

Pass: the valuation date, forecast dates and source periods are explicit.

Test 4 — one-time and reclassification policy

Review restructuring, impairments, litigation, acquisitions, discontinued operations and accounting-policy changes. Do not “normalize” an item simply because it is inconvenient. Record why it is excluded, whether cash has already been paid and whether recurrence is plausible.

Pass: adjustments are visible in a bridge from reported to modeled performance.

Test 5 — link and hard-code control

List external links, pasted values and hard-coded overrides. Inspect formulas for accidental constants. Use a consistent convention for assumptions and formulas, but do not treat color formatting as control evidence by itself.

Pass: no material value depends on an unexplained hard code or broken external link.

Gate 2: accounting-to-cash conversion

Test 6 — cash-flow definition

Confirm what is being valued. FCFF should represent cash available to all capital providers and normally be discounted at a cost of capital. FCFE represents cash available to equity and normally uses cost of equity. A common critical defect is to discount post-debt equity cash flow at WACC or pre-debt firm cash flow at cost of equity.

Pass: the cash-flow definition and discount rate refer to the same claimholders.

Test 7 — non-cash charges

Reconcile depreciation, amortization, stock-based compensation, impairments and other non-cash items. Adding back a non-cash expense does not make it economically free. Related reinvestment, dilution or recurring operating requirements may still matter.

Pass: add-backs have a corresponding economic interpretation.

Test 8 — working-capital mechanics

Model receivables, inventory, payables and other operating balances with appropriate drivers. Confirm signs: growth in operating working capital is usually a cash use. Avoid modeling working capital as an arbitrary percentage of revenue if the business has materially different collection, inventory or supplier dynamics.

Pass: working-capital balances reconcile and produce plausible cash movements.

Test 9 — reinvestment and capital expenditure

Separate maintenance from growth only when evidence supports the distinction. Compare capital expenditure, depreciation, asset age, capacity and growth. A terminal growth story without continuing reinvestment is a warning.

Pass: forecast growth has the operating assets and reinvestment needed to support it.

Test 10 — tax consistency

Distinguish statutory, effective and cash tax rates. Treat tax-loss carryforwards, jurisdiction mix and interest tax effects consistently with the cash-flow method. Do not move from a temporary low effective rate directly to perpetuity without explanation.

Pass: the tax path has a bridge to a sustainable rate and cash basis.

Gate 3: forecast logic

Test 11 — revenue driver bridge

Break revenue into economically meaningful drivers: customers × price, units × price, capacity × utilization, transactions × fee or another business-specific structure. A single growth-rate row hides whether the forecast assumes market-share gain, price, mix or acquisition.

Pass: the model states what must happen operationally for revenue to occur.

Test 12 — margin bridge

Bridge gross, contribution and operating margins through price, mix, productivity, wage, input, acquisition and fixed-cost assumptions. A margin should not expand merely because a historical trend line continues.

Pass: each material margin change has an operational cause and timing.

Test 13 — capacity and headcount

Check whether people, facilities, technology and vendors can deliver the forecast. Look for step costs and implementation delays. Growth without customer support, working capital or distribution capacity may be impossible even if spreadsheet ratios improve.

Pass: resource requirements are modeled or explicitly bounded.

Test 14 — scenario coherence

A downside case should be a coherent state of the business, not every input reduced by 10%. Lower demand can affect price, mix, working capital, capacity and financing differently. Define a causal scenario and make linked assumptions move together.

Pass: a reviewer can explain the economic story of each scenario.

Test 15 — forecast-to-steady-state path

Inspect the final explicit forecast year. Growth, margins, reinvestment, working capital and returns should be moving toward a plausible mature state if the terminal calculation assumes stability. Abrupt changes at the terminal boundary indicate the explicit forecast is too short or poorly designed.

Pass: the final forecast year is a credible starting point for the terminal model.

Gate 4: discount-rate consistency

Test 16 — risk-free rate basis

Match currency and duration. The risk-free input should be in the same currency as the cash flows; location of the company is not enough. Record date and source because rates move.

Pass: the model documents instrument, currency, maturity and observation date.

Test 17 — equity-risk premium and beta

Record whether the equity-risk premium is historical or implied and whether beta is regression-based, bottom-up or adjusted. If using comparable companies, document selection, unlevering, relevering and target capital structure. NYU Stern publishes current data and tools, but using a published number does not eliminate the need to match it to the model.

Pass: ERP and beta are reproducible and consistent with the valuation date.

Test 18 — cost of debt and tax shield

Use a current borrowing-rate estimate or defensible spread, not necessarily the coupon on old debt. Apply the tax shield only when the cash-flow method and tax capacity support it.

Pass: cost of debt reflects current risk and the tax treatment is explained.

Test 19 — capital-structure weights

Use market-value weights where appropriate and specify treatment of leases, pensions, preferred shares, convertibles, minority interests and excess cash. Book-value weights can materially distort WACC.

Pass: the capital bridge is complete and avoids double counting.

Test 20 — nominal/real, pre/post-tax and currency match

Damodaran's valuation examples show that real cash flows discounted at nominal rates, or nominal cash flows discounted at real rates, produce inconsistent values. The same principle applies to pre-tax versus post-tax and currency mismatches.

Pass: every cash-flow basis matches its discount-rate basis.

Gate 5: terminal-value discipline

Test 21 — method choice

State whether terminal value uses perpetual growth, exit multiple or another approach. An exit multiple introduces a pricing assumption into an intrinsic-value model; that can be used as a cross-check, but it should not be disguised.

Pass: method, rationale and limitations are explicit.

Test 22 — sustainable growth

For a perpetual-growth model, test whether long-run growth is plausible for the currency and economy and remains below the discount rate. More importantly, connect growth to reinvestment and return on invested capital. Growth is not free.

Pass: steady-state growth, reinvestment and returns are internally consistent.

Test 23 — terminal margin and returns

Compare terminal margins and returns with business economics, competition and maturity. A company cannot indefinitely increase market share and margins without an explanation of competitive advantage.

Pass: terminal economics are defensible as a mature state.

Test 24 — terminal-value concentration

Calculate terminal value as a percentage of enterprise or equity value. A high share is not automatically wrong, but it means the conclusion depends heavily on long-run assumptions. Disclose the concentration and expand sensitivity or the explicit forecast where useful.

Pass: concentration is measured, explained and reflected in decision confidence.

Test 25 — implied multiple and reverse check

Translate perpetual-growth terminal value into an implied valuation multiple and compare it with the steady-state economics. Also solve for the growth, margin or return assumptions implied by the current price or proposed transaction value.

Pass: the terminal result survives at least one independent economic cross-check.

Gate 6: decision controls

Test 26 — sensitivity design

Use ranges tied to uncertainty, not cosmetically narrow grids. At minimum, test discount rate, terminal growth or multiple and the operating driver that most changes cash flow. Do not let sensitivity substitute for a coherent downside scenario.

Pass: ranges are sourced or justified and include decision-changing values.

Test 27 — bridge to enterprise and equity value

Reconcile present value of operating assets to enterprise value, then equity value. Treat debt, cash, leases, minority interests, non-operating assets, options and diluted shares consistently.

Pass: every bridge item has source, date and sign logic.

Test 28 — independent calculation check

Recalculate a sample of present values, terminal value, WACC and share-price bridge outside the main formula chain. Check timing convention, including mid-year versus year-end discounting.

Pass: independent calculations reproduce the model within explained rounding.

Test 29 — decision memo fidelity

Compare the executive summary with the actual model. Does the memo disclose the range, critical assumptions, open defects and terminal concentration? A balanced model can be misrepresented by a one-number slide.

Pass: the narrative does not imply certainty the model does not contain.

Test 30 — version, access and change control

Protect formula areas, document inputs, separate scenarios, record changes and identify the approved version. If AI tools assisted formula review or assumption research, record the tool, prompt purpose, inputs allowed, output checked and human decision. Do not upload confidential models to an unapproved service.

Pass: another authorized reviewer can identify what changed and reproduce the approved output.

Score the DCF-30 audit

Score each test:

2 = passed with evidence;
1 = partly passed or low-impact gap;
0 = failed or unsupported.

Maximum score is 60. Use this interpretation:

Score	Interpretation	Action
54–60	Strongly controlled	Use with stated scenario and judgment limitations
45–53	Reviewable with conditions	Resolve high defects before material decision
36–44	Weak	Rework and re-audit
Below 36	Unreliable	Do not use for decision

Add two independent gates: no open critical defect and no unresolved mismatch in cash-flow/discount-rate basis. A high total score cannot compensate for a fundamental inconsistency.

Worked audit: a value that is right for the wrong reasons

Consider a fictional DCF with five forecast years:

current revenue: $100 million;
revenue growth: 12%, 10%, 8%, 6%, 5%;
operating margin expands from 14% to 22%;
nominal WACC: 9%;
perpetual growth: 3.5%;
terminal value: 78% of enterprise value;
model reports enterprise value of $420 million.

The workbook calculates correctly. The audit finds four defects.

Defect A — inflation mismatch. Revenue and costs were forecast in “today's money” without inflation, but the WACC is nominal. Test 20 fails critically. The reviewer must either convert cash flows to nominal terms or use a defensible real discount rate.

Defect B — unsupported margin expansion. Revenue drivers are documented, but operating margin rises eight percentage points without price, mix, productivity, headcount or capacity evidence. Tests 12 and 13 fail high.

Defect C — terminal discontinuity. Capital expenditure falls below depreciation in the final forecast year even though perpetual growth is 3.5%. Reinvestment is inconsistent with the steady-state story. Tests 9, 15 and 22 fail.

Defect D — presentation bias. The memo states “fair value $420 million” but does not disclose that terminal value contributes 78% or that a one-point change in WACC alters the decision. Tests 24 and 29 fail.

Suppose the initial score is 41/60. Correcting only the formulas is not enough. The owner rebuilds nominal forecasts, ties margin improvement to documented operating drivers, restores steady-state reinvestment and presents a range of $310–$405 million. The revised base value may be lower, but the model is more useful because the decision makers can see what would have to be true.

A 15-minute first-pass review

If time is limited, do these five checks first:

Identify FCFF or FCFE and match the discount rate.
Trace the last actual cash-flow calculation to statements.
calculate terminal value as a share of total value.
Match nominal/real, currency and tax bases.
Find the one operating assumption that moves the recommendation most.

If any of the first four fail, stop interpreting the headline value. Continue the full audit after the basis is repaired.

How to present a DCF portfolio project

A credible learning artifact should include:

a one-page company and decision context;
assumptions register with sources and dates;
historical-to-forecast bridge;
explicit cash-flow definition;
base, downside and upside scenarios;
WACC build with source notes;
terminal-value checks;
DCF-30 score and defect log;
executive recommendation with limitations;
change log showing how review altered the result.

Do not publish proprietary company data, confidential transaction details or unlicensed paid-source extracts. A fictional case or a model built from public filings is safer for a portfolio, provided every public source is attributed and the work is clearly your analysis.

Go deeper with a relevant programme

The Executive Certificate in Strategic Finance, M&A & Corporate Valuation is the relevant MTF pathway for learners who want structured practice with financial analysis, valuation, capital allocation and deal decisions. Use DCF-30 to evaluate your own model before treating a completed spreadsheet as evidence of capability.

Sources

Prepared on 17 September 2026. This is an educational review framework, not investment, accounting, tax, legal or valuation advice. Real decisions require appropriately qualified review and current evidence.