DCF Valuation Explained: How to Value a Company Step by Step

A discounted cash flow (DCF) valuation estimates what a company is worth today by projecting its future free cash flows and discounting them back to the present using its weighted average cost of capital (WACC). It is the only valuation method that ties price to economics rather than to what comparable companies happen to trade for on a given Tuesday. This guide walks through the full DCF model — projection, WACC, terminal value, sensitivity — at the level a banker, PE associate, or operator actually needs to defend in a room.

Used well, a DCF is the backbone of every fairness opinion filed in an SEC merger proxy. Microsoft's $68.7 billion acquisition of Activision Blizzard, announced January 18, 2022 at $95 per share, was tested with a standalone DCF that produced an implied value of roughly $97 per share — close enough to the deal price that the board could defend it as fair. Get the mechanics right and you get a valuation a board can sign. Get them wrong and you're handing a plaintiff's lawyer a roadmap.

What a DCF Actually Measures

A DCF answers one question: what is the present value of all the cash a business will generate for its capital providers, from today until the end of time? Aswath Damodaran, the NYU Stern finance professor whose valuation data sets are the de facto reference for the industry, frames it as the sum of two pieces — an explicit forecast period (usually 5–10 years) and a terminal value capturing everything after that.

The structure is simple. The discipline lives in the inputs:

Free cash flow to the firm (FCFF) — EBIT × (1 – tax rate) + depreciation & amortization – capital expenditure – change in working capital. This is the cash available to both debt and equity holders.
Weighted average cost of capital (WACC) — the blended required return across the firm's debt and equity capital.
Terminal value — the lump-sum estimate of all cash flows beyond the forecast horizon.
Net debt bridge — to move from enterprise value to equity value, subtract debt and add cash.

Takeaway: Before you open a spreadsheet, write down the four inputs above on paper. If you can't defend each one with a public benchmark or a management forecast, the model is decoration.

Step 1: Project Free Cash Flow (Years 1–5)

The forecast period is where most DCFs go wrong — analysts extrapolate revenue growth straight off a hockey stick and never connect it to operating economics. McKinsey's Valuation: Measuring and Managing the Value of Companies, now in its eighth edition (2024) and used as the standard text in most MBA corporate finance programs, anchors its framework on a different principle: only two things drive value — return on invested capital (ROIC) and growth. Anything you forecast must trace back to those two.

A defensible 5-year projection works in this order:

Revenue — build it bottom-up from price × volume by segment, not top-down from "we'll grow 20%". Tie growth to a public benchmark: industry CAGR, comparable peer growth, or management's investor day guidance.
EBIT margin — assume convergence toward peer median over the forecast period. Salesforce's operating margin took roughly a decade to converge with mature enterprise software peers; don't assume your subject does it in two.
Tax rate — use the marginal statutory rate (21% federal for U.S. companies post-TCJA, plus state) rather than the effective rate, which is distorted by one-time items.
Reinvestment — capex plus change in working capital, calibrated so that EBIT growth ÷ reinvestment yields a sensible ROIC.

Warren Buffett's "owner earnings" concept, introduced in his 1986 Berkshire Hathaway shareholder letter, is a useful sanity check on the FCF line: reported earnings + non-cash charges – the capex needed to maintain unit volume and competitive position. If your forecast FCF runs persistently above owner earnings, you're likely under-counting maintenance capex.

Takeaway: Every line in your forecast should have a comment cell citing the source (filings, broker estimates, management guidance, industry report). If a reviewer can't audit each number to a primary source in under 60 seconds, the model isn't review-ready.

Step 2: Calculate WACC

WACC is the discount rate. It's the blended cost of the firm's capital, weighted by the market value of debt and equity:

WACC = (E/V) × Cost of Equity + (D/V) × Cost of Debt × (1 – Tax Rate)

Cost of equity uses the Capital Asset Pricing Model: Risk-free rate + Beta × Equity Risk Premium. Each input needs to come from a defensible public source:

Risk-free rate — the 10-year U.S. Treasury yield on the valuation date. Use the actual current yield, not a historical average.
Equity risk premium (ERP) — Damodaran's implied ERP for the U.S. market was approximately 4.4% as of early 2025, published in his January 2025 valuation packet. Most fairness opinions use a range of 5–6% based on Duff & Phelps / Kroll surveys; pick a methodology and stay consistent.
Beta — calculate as the slope of weekly returns vs. the S&P 500 over 2–5 years, or pull a re-levered industry beta from Damodaran's free industry data set on the NYU Stern site.
Cost of debt — the yield to maturity on the company's outstanding long-dated debt, or for a private company, the spread for its credit rating off the risk-free rate.
Capital structure weights — use the firm's target capital structure or the industry average, not the snapshot from today's balance sheet, which can be distorted by one-time issuances or buybacks.

A 1% error in WACC moves enterprise value by roughly 10–20% for a typical mature company. This is the most sensitive input in the entire model.

Takeaway: Sensitivity-test WACC ±100 basis points and ±200 basis points before you ever show a number to a decision-maker. A point estimate without a range is malpractice.

Step 3: Build Terminal Value

The terminal value typically accounts for around 75% of the total enterprise value in a DCF, according to Wall Street Prep's practitioner reference — so getting it wrong overwhelms everything you did in the projection period. Two methods, used together:

Perpetuity Growth (Gordon Growth) Method

TV = FCF_{final year} × (1 + g) / (WACC – g)

The growth rate g must be below long-run nominal GDP growth, which has historically averaged around 4% for the U.S. (roughly 2% real plus 2% inflation). A perpetuity growth rate above GDP implies the company eventually becomes larger than the economy — a flag every investment committee will spot. Common defensible ranges: 2.0% to 3.0% for developed-market companies.

Exit Multiple Method

TV = EBITDA_{final year} × Exit Multiple

The exit multiple should be drawn from current trading multiples of mature peers, not from the subject company's current multiple (which often embeds growth premia that won't exist at terminal). Microsoft's Activision deal closed at an implied EV/LTM EBITDA of roughly 20.8x — above the 14–18x trading range for comparable game publishers because it embedded a strategic control premium. In a terminal value, you want the steady-state multiple, not the M&A premium.

Best practice — used by every credible fairness opinion — is to run both methods and triangulate. If your perpetuity growth method implies a 30x exit multiple while comps trade at 12x, one of the two methods has a bad assumption. The numbers should land within a reasonable band.

Takeaway: Always reverse-engineer the implied exit multiple from your perpetuity calculation, and vice versa. They are the same number expressed two ways; if they disagree, your model is internally inconsistent.

Step 4: Discount, Bridge, and Sensitize

Once you have annual FCFs and a terminal value, discount each cash flow back to today using the WACC:

Enterprise Value = Σ [FCF_t / (1+WACC)^t] + TV / (1+WACC)ⁿ

Use the mid-year convention (discount each year's FCF as if it arrives mid-year) for any operating business with smooth cash generation — it adds roughly 0.5 × WACC of value vs. end-of-year discounting and is standard in banking models.

Then build the equity bridge:

Enterprise Value
– Total Debt
– Preferred Equity, Minority Interest, Unfunded Pension
+ Cash & Marketable Securities
= Equity Value
÷ Diluted Shares Outstanding (treasury stock method)
= Implied Share Price

Then sensitize. Every defensible DCF presents a 2-D output table showing enterprise value across a WACC range (typically ±100 bps) and a terminal value driver range (perpetuity growth ±50 bps, or exit multiple ±2 turns). Morningstar's published Activision Blizzard fair value of $92 per share during the Microsoft deal period was triangulated against the $95 offer using exactly this kind of band — a single point estimate would have been useless to investors weighing whether the deal would close.

Takeaway: Never present a DCF as a single number. Present it as a range, with explicit ranges on the two most sensitive drivers, and label the assumptions behind the midpoint.

Step 5: Stress-Test the Model Like a Skeptical Buyer

The final step separates a DCF you can defend from one that gets torn apart in committee. Before you present, run these checks:

Implied ROIC at terminal — does the steady-state return on invested capital exceed WACC by a sensible spread (typically 2–5%)? If ROIC = WACC, growth creates zero value and the perpetuity growth rate should be zero.
Terminal value as % of EV — if it's above 80%, your forecast period is too short or your growth fade is too gentle. Extend to 10 years and re-run.
Implied entry multiple vs. comps — back out the implied EV/EBITDA and EV/Revenue from your DCF and overlay them on trading and transaction comps. A DCF that produces a multiple miles outside the comp set needs an explicit narrative.
Reinvestment consistency — growth must be funded by reinvestment. If you're forecasting 8% revenue growth with capex at maintenance levels, the model is internally inconsistent.
Margin reversion — if margins are at cyclical peaks, fade them toward mid-cycle. McKinsey's research on cyclical companies makes this point explicitly: valuing a cyclical at peak earnings produces nonsense.

Takeaway: Build a one-page "model integrity check" tab showing each of the five checks above with a pass/fail flag. Every M&A model that ends up in a board deck has one — yours should too.

From Spreadsheet to Decision

A DCF is not a price tag — it's a structured argument. The argument has roughly thirty inputs, each of which can be defended or challenged. When Microsoft's board approved the Activision deal at $95, they had banker DCFs ranging from the high-$80s to the low-$100s, exit multiples ranging from 18x to 24x, and WACCs in a 200-bps band. The decision wasn't "the DCF says $95" — it was "across every reasonable set of assumptions, $95 sits inside the defensible range."

Building that model from scratch in Excel takes a senior associate roughly 8–12 hours of clean work, plus another 4–6 hours of formatting and review. Most operators and founders neither have those hours nor want to rebuild WACC schedules from first principles every time they value an acquisition target or test their own equity story. A pre-built DCF template with a hard-coded WACC schedule, mid-year convention toggle, dual terminal value methodology, and integrated sensitivity tables turns a multi-day modeling exercise into a multi-hour input exercise. That's the difference between a model you finish on a Friday and a model you skip on a Tuesday.