DCF Modeling: A Step-by-Step Guide

A discounted cash flow (DCF) analysis values a company as the present value of the cash it is expected to generate for its capital providers. It is the most cited intrinsic valuation method in investment banking interviews because it forces you to defend a complete set of operating, financing, and growth assumptions rather than point at where comparable companies trade.

This guide builds a DCF the way it is built on a deal team and the way it is taught in Aswath Damodaran's valuation work at NYU Stern and the CFA Institute equity valuation curriculum. You will project unlevered free cash flow, derive a discount rate, estimate terminal value, bridge to equity value per share, and pressure-test the output. One worked example with consistent numbers runs through every section so you can follow the arithmetic end to end.

What is a DCF?

A DCF estimates intrinsic value by discounting a company's projected future free cash flows back to today at a rate that reflects their risk. The governing idea is the time value of money: a dollar received in five years is worth less than a dollar today because today's dollar can be invested and because future cash is uncertain.

The DCF is an intrinsic method. It values a business on its own forecast cash flows and risk profile, independent of current market sentiment. This is its main contrast with relative valuation (comparable companies and precedent transactions), which prices a company off multiples paid for similar businesses. In a live valuation you triangulate: the DCF anchors fundamental value, and trading and transaction comps tell you what the market and acquirers are actually paying.

The standard structure

A banking DCF almost always values the firm using unlevered free cash flow (free cash flow to the firm), not levered free cash flow to equity. Unlevered cash flow is the cash the operating business produces before any financing decisions. Discounting it at the weighted average cost of capital (WACC) produces enterprise value. You then subtract net debt and other non-equity claims to reach equity value, and divide by diluted shares for value per share.

The full sequence is:

1. Project unlevered free cash flow for an explicit forecast period, usually 5 to 10 years. 2. Estimate a terminal value capturing all cash flows beyond the forecast period. 3. Discount the forecast cash flows and the terminal value to the present at WACC. 4. Sum them to get enterprise value. 5. Bridge from enterprise value to equity value, then to per-share value. 6. Sensitize the key assumptions to produce a value range.

Working with unlevered cash flow and WACC keeps the operating and financing decisions separate, which is cleaner than a levered (free cash flow to equity, discounted at cost of equity) model when capital structure is expected to change.

When a DCF is the right tool

A DCF is most reliable when cash flows are reasonably predictable and the company is at or near a steady operating state: mature industrials, consumer staples, regulated utilities, established software with durable retention. It is weakest for early-stage companies with negative cash flow, deeply cyclical businesses valued at the wrong point in the cycle, financial institutions (where the unlevered framework breaks down and you value equity directly), and any situation where terminal value assumptions swamp the explicit forecast.

In those cases comparable companies and precedent transactions carry more weight. Comps reflect what the market pays for similar growth and margin profiles right now; precedent transactions capture control premiums actually paid in M&A. The honest interview answer is that no single method is definitive. The DCF gives you a fundamentals-based number and, just as important, an explicit list of the assumptions that drive it.

DCF versus relative valuation

The three core valuation methods answer different questions, and a banker uses them together to build a defensible range.

In a "football field" valuation summary, the DCF range is plotted next to the comps and precedent ranges. Convergence across methods builds confidence; a wide gap is a signal to revisit assumptions or question the market. A frequent interview question is which method gives the highest value. Precedent transactions usually run highest because they embed a control premium, while trading comps reflect a minority-stake market price. The DCF can land anywhere depending on assumptions, which is exactly why it is sensitized rather than quoted as a single number.

The headline criticism

The standard criticism of the DCF, and one Damodaran returns to often, is that it produces a precise-looking number from imprecise inputs. Small changes in the discount rate or terminal growth rate move the output materially, and terminal value typically accounts for the majority of total value. A DCF is only as good as its assumptions, so the discipline is in defending each one and showing the range, not in the false comfort of a single point estimate.

A second criticism is the circularity in the discount rate. WACC uses the market value of equity, which is what the DCF is trying to estimate in the first place. In practice you use the current market capitalization or a target capital structure and accept the approximation. A third is forecast horizon risk: pushing growth too far into the explicit period or letting the final year settle at a non-steady-state margin distorts the terminal value, which then propagates through most of the answer. None of these flaws disqualify the DCF. They define where the analytical effort and the interview scrutiny belong.

Projecting Free Cash Flows

Unlevered free cash flow equals EBIT times (1 minus the tax rate), plus depreciation and amortization, minus capital expenditures, minus the increase in net working capital.

Written out:

Unlevered FCF = EBIT x (1 - tax rate) + D&A - CapEx - Increase in NWC

EBIT times (1 minus the tax rate) is often called NOPAT, net operating profit after tax. Each term has a clear rationale:

• EBIT x (1 - tax rate): Start from operating profit and tax it as if the firm had no debt. Using EBIT (not net income) keeps interest, a financing item, out of the cash flow. The tax is a notional unlevered tax so the financing decision does not contaminate operating cash flow.
• Plus D&A: Depreciation and amortization reduced EBIT but moved no cash, so add them back.
• Minus CapEx: Capital expenditures are real cash outflows that sustain and grow the asset base but do not run through EBIT.
• Minus increase in net working capital: Growth ties up cash in receivables and inventory net of payables. An increase in NWC consumes cash; a decrease releases it.

This is unlevered because it is calculated before interest. The same cash flow is available to all capital providers, debt and equity, which is why it is discounted at WACC and yields enterprise value.

Building the projection

Forecasts typically span 5 to 10 years. The period should be long enough that the company reaches a stable margin and reinvestment profile by the final year, because the terminal value assumes steady state from that point. Drive each line from explicit assumptions:

• Revenue: Build from volume and price, or growth rates anchored to historical trends, management guidance, and industry forecasts. Growth should decay toward a sustainable long-run rate by the final year.
• EBIT margin: Project operating margin from historicals and an operating leverage view. Avoid heroic margin expansion you cannot defend.
• Tax rate: Use a normalized marginal rate, often the statutory rate, rather than a volatile effective rate distorted by one-time items.
• D&A and CapEx: Tie both to revenue or the asset base. In steady state, CapEx should modestly exceed D&A for a growing company; they converge as growth slows.
• Net working capital: Project from days metrics (days sales outstanding, days inventory, days payable) or as a percentage of revenue, then take the year-over-year change.

In an interview, expect to defend every assumption. The most common follow-up is "why that growth rate" or "why that margin," so each input needs a one-sentence justification grounded in history, guidance, or industry economics.

Unlevered versus levered cash flow

A frequent interview trap is the difference between unlevered free cash flow (free cash flow to the firm) and levered free cash flow (free cash flow to equity). Unlevered cash flow excludes interest entirely and is discounted at WACC to produce enterprise value. Levered cash flow subtracts interest expense and mandatory debt repayments, represents cash available only to equity holders, and is discounted at the cost of equity to produce equity value directly. The two routes should reconcile to a similar equity value when assumptions are internally consistent. Bankers default to the unlevered approach because it separates operating performance from financing policy, which keeps the valuation stable even when the capital structure is expected to change over the forecast.

Common projection mistakes

Three errors recur in candidate models and in interview answers. First, double-counting or omitting the change in net working capital: an increase in NWC is a use of cash and must be subtracted, and the sign is easy to flip under pressure. Second, letting CapEx and D&A drift apart in the terminal year. A company growing at a low single-digit rate forever cannot keep spending far more on CapEx than it depreciates; the two should roughly converge by the final year so the steady state is internally consistent. Third, taxing net income instead of EBIT. The unlevered framework requires a notional tax on operating profit, before any interest benefit, so the financing decision does not leak into operating cash flow. Getting these three right signals real modeling fluency.

Worked example: the projection

Consider a mature company with Year 1 EBIT of $1,000 and a flat 25% tax rate. EBIT grows 6% in Years 2 and 3, then 4% in Years 4 and 5. D&A runs at $200 in Year 1 growing 4% a year, CapEx at $250 in Year 1 growing 4% a year, and the increase in net working capital is $50 a year held flat for simplicity.

These five unlevered free cash flows, 650, 693, 739, 771, and 804, carry through the rest of the model.

Calculating WACC

The weighted average cost of capital is the blended, after-tax required return of a firm's debt and equity providers, weighted by the market value of each. It is the discount rate applied to unlevered free cash flow.

The formula is:

WACC = (E/V) x Re + (D/V) x Rd x (1 - tax rate)

where E is the market value of equity, D is the market value of debt, V is E plus D, Re is the cost of equity, Rd is the pre-tax cost of debt, and the tax rate captures the tax deductibility of interest.

Two points bankers get tested on. First, use market values for the weights, not book values. Equity weight uses market capitalization; debt uses market value of debt (book value is an acceptable proxy when debt is not actively traded). Second, the cost of debt is multiplied by (1 minus the tax rate) because interest is tax deductible, which lowers the effective cost of borrowing. There is no equivalent shield on equity.

Cost of equity via CAPM

The cost of equity is estimated with the capital asset pricing model:

Re = Rf + Beta x (equity risk premium)

• Risk-free rate (Rf): The yield on a long-dated government bond, typically the 10-year US Treasury, matched to the long horizon of the cash flows.
• Beta: The sensitivity of the stock's returns to the market. Source a comparable-company set, unlever each peer's beta, take the median, then relever at the target capital structure. This produces a beta consistent with the company's own leverage rather than a single noisy regression beta.
• Equity risk premium (ERP): The extra return investors require for holding equities over the risk-free asset.

On the equity risk premium, use a current, defensible number rather than the dated textbook "5 to 7%." Damodaran's implied equity risk premium work, which backs the premium out of current index prices and expected cash flows, has pointed to a mature-market premium in roughly the 4.5% to 5.5% range in recent years. The CFA Institute equity valuation curriculum similarly treats the ERP as an estimated, time-varying input rather than a fixed constant. Pick a point in that range and be ready to justify it.

On beta specifically, the unlever-then-relever process matters and is heavily tested. Each comparable's observed (levered) beta reflects both its business risk and its financial leverage. Strip out leverage with the Hamada relationship, take the median asset (unlevered) beta across the peer set to isolate pure business risk, then relever at the target company's own debt-to-equity ratio. This produces a beta consistent with the company's capital structure rather than a single noisy regression on its own stock, which is the approach Damodaran's data-driven valuation work and the CFA Institute curriculum both favor.

Cost of debt

The pre-tax cost of debt is the yield the company would pay on new borrowing today. Estimate it from the yield to maturity on existing traded debt, or from a synthetic rating: map an interest coverage ratio to a credit rating, add the corresponding default spread to the risk-free rate. Then apply the (1 minus tax rate) shield.

Use the yield, not the coupon. A bond's coupon reflects rates when it was issued; the yield to maturity reflects what the market demands today, which is the relevant marginal cost of new debt. For a healthy investment-grade company the cost of debt sits well below the cost of equity because debtholders rank ahead of equity in the capital structure and bear less risk, and because of the interest tax shield. That ordering, equity always costing more than debt for the same company, is a common interview check.

Why WACC, not cost of equity

Unlevered free cash flow is available to every capital provider, so it must be discounted at the blended required return of every capital provider, which is WACC. Discounting unlevered cash flow at the cost of equity would mismatch the numerator (firm-wide cash) with the denominator (an equity-only rate) and overstate the discount, understating value. The levered alternative, free cash flow to equity discounted at the cost of equity to reach equity value directly, is internally consistent but more fragile when the capital structure is expected to change, which is why the unlevered-WACC framework is the banking default.

Worked example: WACC

Continue the example. Assume:

• Risk-free rate: 4.0%
• Relevered beta: 1.10
• Equity risk premium: 5.0% (within the Damodaran implied range)
• Pre-tax cost of debt: 6.0%
• Tax rate: 25%
• Capital structure at market value: 80% equity, 20% debt

Cost of equity: Re = 4.0% + 1.10 x 5.0% = 4.0% + 5.5% = 9.5%

After-tax cost of debt: 6.0% x (1 - 0.25) = 4.5%

WACC = 0.80 x 9.5% + 0.20 x 4.5% = 7.6% + 0.9% = 8.5%

The model uses a WACC of 8.5% to discount the unlevered free cash flows and the terminal value.

Terminal Value Methods

Terminal value captures the value of all cash flows beyond the explicit forecast period, since a company is assumed to operate indefinitely but cannot be modeled year by year forever. There are two standard methods, and a disciplined DCF computes both and checks them against each other.

Gordon Growth (perpetuity growth) method

The Gordon Growth method treats final-year cash flow as growing at a constant rate forever:

Terminal Value = FCF(n+1) / (WACC - g)

FCF(n+1) is the first cash flow after the forecast period, equal to final-year FCF times (1 + g). The perpetual growth rate g must be pinned to a sustainable long-run rate, roughly long-run GDP growth or inflation, on the order of 2% to 3%. A higher g implies the company eventually grows faster than the economy forever, which is impossible. The arithmetic also breaks down as g approaches WACC, another reason to keep g conservative.

Exit multiple method

The exit multiple method applies a valuation multiple, most commonly EV/EBITDA, to the final-year operating metric:

Terminal Value = Final-year EBITDA x Exit EV/EBITDA multiple

The multiple should reflect where a mature company in the sector would reasonably trade at the end of the forecast, typically informed by current trading comps with a haircut for maturity. The exit multiple method is market-based and intuitive but imports current multiple conditions into the future, so it should be cross-checked against the implied perpetual growth rate it produces.

Best practice is to compute terminal value both ways and confirm they imply consistent economics. If the exit multiple implies a 6% perpetual growth rate, the multiple is too high.

To run that cross-check, take the terminal value from the exit multiple method and solve the Gordon Growth formula backward for the implied g: rearranging Terminal Value = FCF(n+1) / (WACC - g) gives an implied growth rate. If it is below long-run GDP, the multiple may be conservative; if it exceeds GDP by a wide margin, the multiple is unsustainable and should come down. Running the same check in reverse, taking the Gordon Growth terminal value and computing the implied EV/EBITDA exit multiple, tells you whether your perpetual growth assumption translates into a multiple the market would actually pay. Reconciling the two methods is the single most reliable way to keep terminal value honest.

Discounting and the weight of terminal value

Terminal value is a value as of the end of the final forecast year, so it must be discounted back to the present at the same rate as the final-year cash flow.

Terminal value typically represents 60% to 80% of total enterprise value in a DCF. That concentration is precisely why interviewers probe terminal assumptions hardest. If most of your value sits in a single perpetuity formula, the growth rate and discount rate in that formula deserve the most scrutiny.

Mid-year convention

A refinement worth mentioning: the mid-year convention. The base model implicitly assumes all cash arrives on the last day of each year. In reality cash flows in throughout the year, so discounting at full-period factors understates value slightly. The mid-year convention discounts each year's flow as if received at mid-year (period 0.5, 1.5, 2.5, and so on), modestly increasing present value. It is a standard adjustment and a common interview detail. For clarity, the worked example below uses full-year (year-end) discounting.

Worked example: terminal value

Use the Gordon Growth method with a perpetual growth rate of 2.5%. Year 5 unlevered FCF is 804, and WACC is 8.5%.

FCF(6) = 804 x (1 + 0.025) = 824

Terminal Value (end of Year 5) = 824 / (0.085 - 0.025) = 824 / 0.060 = 13,733

This terminal value sits at the end of Year 5 and will be discounted back five years alongside the Year 5 cash flow.

Sensitivity Analysis

A DCF is highly sensitive to a small number of assumptions, principally WACC and the terminal growth rate, so the output should be presented as a range produced by a sensitivity table rather than a single number. This is the final step and the one that signals you understand the model's limitations.

Completing the worked example

First, discount each year's unlevered free cash flow at 8.5% (year-end convention):

Sum of present value of forecast FCFs = 599 + 589 + 579 + 556 + 535 = 2,858

Present value of terminal value = 13,733 x 0.6650 = 9,132

Enterprise value = 2,858 + 9,132 = 11,990

Notice the terminal value contributes 9,132 of 11,990, about 76% of enterprise value, squarely in the typical 60% to 80% range.

The bridge to equity value per share

Enterprise value belongs to all capital providers. To reach equity value, subtract net debt and any other non-equity claims:

Equity Value = Enterprise Value - Net Debt - Other Claims

Net debt is total debt minus cash and equivalents. Other claims can include preferred stock, minority (noncontrolling) interest, and unfunded pension obligations, added or subtracted depending on the item.

Assume net debt of 2,000 and no other claims:

Equity value = 11,990 - 2,000 = 9,990

Divide by diluted shares outstanding to get per-share value. Diluted shares are computed using the treasury stock method, which assumes in-the-money options and warrants are exercised and the proceeds are used to repurchase shares at the current price, capturing only the net dilution.

Assume 500 diluted shares (treasury stock method):

Value per share = 9,990 / 500 = $19.98

That $19.98 is the DCF's intrinsic per-share estimate, which you would then compare to the current trading price and to the comps and precedent transaction ranges.

Building the sensitivity table

Because WACC and terminal growth drive most of the value, the standard output is a two-variable data table flexing both. Holding the example's structure, varying WACC across 7.5%, 8.5%, and 9.5% and g across 2.0%, 2.5%, and 3.0% produces a grid of per-share values that widens as WACC falls and g rises:

(Cells are illustrative of the spread, computed on the same cash flow stream.) The base case of WACC 8.5% and g 2.5% sits near the center at roughly $20. The honest takeaway is a defensible range, not a single figure, with the per-share value moving by more than 50% across reasonable input choices.

What sensitivities to run, and why it matters

Beyond WACC and terminal growth, sensitize the assumptions specific to the company's value drivers: revenue growth, EBIT margin, and the exit multiple if you used the multiple method. The point is not to hide behind a range but to show which assumptions actually move the answer so the discussion focuses there.

In an interview, always volunteer that you would sensitize the key inputs and present a range. It demonstrates that you understand the central criticism of the DCF, that terminal value dominates and the output is only as precise as the inputs, and that you treat the model as a disciplined framework for reasoning about value rather than a black box that prints a price.