01What a DCF model actually says
A company is worth the cash it will hand to its owners over its life, counted in today's money. That is the whole idea. A DCF makes it operational in three moves: forecast free cash flow for a handful of years, discount each year back to today, and compress everything beyond the forecast into a terminal value.
Free cash flow is operating cash flow minus capital expenditure: what is left after the business has paid to keep itself running. Not EBITDA, not net profit. EBITDA ignores the capex a capital-intensive company cannot skip, and net profit is an accounting figure that can move without a cent changing hands.
Discounting is dividing each year's cash flow by (1 + rate) raised to the number of years. A euro in year five at a 9% rate is worth 0.65 today. The rate is the WACC, the weighted cost of the company's debt and equity, because that is what the cash has to earn to be worth having.
02Building one, step by step
- 1. Start from last year's free cash flow, taken from the cash flow statement, not from a broker summary. Operating cash flow minus capex. If the last year was abnormal (a one-off sale, a pandemic year), use a normalised figure and say so.
- 2. Choose a growth rate for the explicit years, five or ten. It should be defensible from the company's history and its market, not from what makes the result look good. Double-digit growth for ten years is a claim, not an input.
- 3. Choose the discount rate. A WACC between 7% and 10% covers most large listed companies; a riskier or smaller business sits higher. If you do not have one, the WACC calculator builds it from the cost of equity, the cost of debt and the capital structure.
- 4. Choose perpetual growth for the terminal value: the rate the company can grow forever. It cannot exceed the long-run growth of the economy, so 2% to 3% is the honest ceiling. Anything higher means the company eventually becomes the economy.
- 5. Add it up, subtract net debt, divide by shares. Then, and this is the step most models skip, change the discount rate by half a point and perpetual growth by half a point and look at how far the result moves.
Do it in the browser: DCF calculator with a 25-scenario sensitivity grid →
03A worked example you can redo by hand
Free cash flow of €100M this year, growing 5% a year for five years (€105M in year one, €127.6M in year five), a WACC of 9% and perpetual growth of 2%. The five projected years, discounted, add up to about €450M. Year six cash flow is €127.6M × 1.02, about €130M; divided by (9% − 2%) that is a terminal value of about €1,860M, which discounted five years back to today leaves about €1,210M. Enterprise value: roughly €1,660M.
Now look at the split. The terminal value is 73% of the total. Five years of careful forecasting decided little more than a quarter of the answer; one assumption about year six onwards decided the rest. That is not a flaw of this example. It is what a DCF is: mostly a statement about the far future, dressed as a calculation about the near one.
The figures are invented and round on purpose, so the arithmetic can be checked with a pencil. They do not belong to any real company.
04Where it breaks: the three inputs that own the result
Terminal value. It typically weighs 70% to 80% of the total, and its formula divides by (WACC − g). As the two rates get closer, the denominator shrinks and the value explodes; if g reaches the WACC, the value is infinite and the model has stopped meaning anything. A terminal weight above 80% is a signal to distrust the result, not to celebrate it.
The discount rate. Half a point of WACC moves the value per share by between 7% and 16% depending on how much debt sits between enterprise value and equity, and more when WACC and perpetual growth are close: in the example above, with no debt, it is 7% to 8%. Nobody knows a company's cost of capital to the half point. A model that returns «€47.32 per share» is claiming a precision its inputs do not have. The sensitivity of a DCF to the WACC deserves its own article, and has one.
The starting cash flow. Every later year is a multiple of year one, so an error there is multiplied through the whole model. The usual mistakes: taking EBITDA as if it were cash, forgetting that working capital eats cash when a company grows, and starting from a record year as if it were normal. Reading the cash flow statement for three or four years before choosing the starting point is the cheapest insurance a valuation can buy.
05How to read a DCF someone else built
- Ask for the terminal weight. If it is not shown, compute it. Above 80% the valuation is a bet on the perpetual growth rate, whatever the explicit years say.
- Ask for the range, not the figure. A serious model shows what happens at WACC ± 0.5 and g ± 0.5. If the range runs from half to double, the honest conclusion is that the method does not conclude for that company.
- Check the starting cash flow against the filed accounts. The cash flow statement is public. If the model's year zero does not match operating cash flow minus capex in the annual report, everything downstream is built on sand.
- Compare the implied growth with the company's history. If the model needs ten years of 15% growth and the company has delivered 4%, the model is describing a different company.
- Reverse it. Start from today's price and solve for the growth the market is pricing in. If that implied growth is modest, the stock may be cheap; if it is heroic, you know what you are being asked to believe. Celsmar's valuation module runs this reverse DCF on filed accounts.
06What a DCF cannot do
It cannot value a company with no foreseeable positive cash flow: a loss-making start-up, a bank (where debt is the product, not the financing) or an exploration company whose cash depends on one discovery. For those, multiples such as EV/EBITDA or P/E against comparable companies say more, with their own limits. And it cannot predict: the model tells you what the company is worth if the assumptions hold. Whether they hold is the whole job.
The three-stage variant, with a high-growth phase, a fade and a mature phase, fixes one problem (a constant growth rate for ten years is rarely realistic) and does not touch the other two: the terminal value still dominates and the discount rate still moves the answer. The three-stage DCF entry in the glossary explains when it is worth the extra complexity.
Frequently asked questions
What is a DCF model?
A discounted cash flow model estimates what a company is worth today from the cash it is expected to generate: you forecast free cash flow for several years, discount each year back to today by dividing by (1 + WACC) raised to the number of years, add a terminal value for everything after the forecast, and sum it all. Subtracting net debt and dividing by the share count gives the value per share.
Why does the terminal value weigh so much in a DCF?
Because it condenses an infinite horizon into one figure, against five or ten explicitly forecast years. It typically accounts for 70% to 80% of the total value. Its formula divides year n+1 cash flow by (WACC − perpetual growth), so when the two rates are close the terminal value becomes very large, and if perpetual growth equals or exceeds the WACC it becomes infinite and the model stops making sense.
What discount rate should a DCF use?
The company's weighted average cost of capital (WACC): the cost of equity and the after-tax cost of debt, weighted by their share of the capital structure. For most large listed companies it falls between 7% and 10%; smaller or riskier businesses sit higher. Because half a point of WACC moves the value per share by 7% to 16% depending on leverage, the result should always be shown as a range across several rates rather than as one figure.
Is a DCF valuation reliable?
As arithmetic, yes; as a prediction, no. A DCF is exact calculation over uncertain assumptions, and its result inherits that uncertainty. It is most useful run in reverse: starting from today's price and solving for the growth the market is implying, which tells you what you would have to believe to buy at that price.
The figures in the worked example are invented and round, so the arithmetic can be redone by hand (year zero €100M, growth 5%, WACC 9%, perpetual growth 2%). They do not correspond to any real company. The 7% to 16% sensitivity to half a point of WACC comes from the grid of the DCF calculator on this site across several capital structures; the example itself, with no debt, moves 7% to 8%.
This is information and analysis, not financial advice, and it contains no recommendation to buy or sell. See the disclaimer.