What a company is worth, and how little certainty a DCF gives you.
Forecast the cash, build the discount rate from its parts, and put a value on everything after the forecast ends. Then watch the answer move by half its own size when you nudge two assumptions, which is the part most tutorials leave out.
How the model works
Free cash flow
Operating profit after tax, add back the non-cash depreciation charge, subtract what
is actually spent on assets and working capital.
FCF = EBIT(1−t) + D&A − CapEx − ΔNWC
Cost of equity
The risk-free rate plus a premium scaled by how much the shares move with the market.
Re = Rf + β × ERP
WACC
Blended by how much of the capital each side provides, with debt taken after tax
because interest is deductible.
WACC = We·Re + Wd·Rd(1−t)
Terminal value
Either a growing perpetuity or a multiple of final year EBITDA.
TV = FCFN(1+g) / (WACC − g)
Growth fade
Growth moves in a straight line from the first forecast year to the last, so a high starting rate is not applied unchanged for a decade.
To value per share
Discount everything, subtract net debt, divide by shares.
Equity = Σ PV(FCF) + PV(TV) − net debt
What the model shows
A discounted cash flow model has a reputation for rigor that it does not entirely deserve. The arithmetic is exact. The inputs are guesses. Understanding which parts are which is most of the skill.
What discounted cash flow actually means
A business is worth the cash it will hand over, and a dollar arriving in year eight is worth less than one arriving today. So: forecast the cash, decide what rate makes future cash comparable to present cash, and add it up. Everything else is bookkeeping around those two steps.
The cash figure is unlevered free cash flow, meaning before interest, so it belongs to lenders and shareholders together. That is why discounting it produces enterprise value, the worth of the operating business, and why you subtract net debt at the end to get what is left for shareholders.
Those two steps together are a present value calculation, and the arithmetic is the same one capital budgeting calls net present value (NPV). The difference is only in what gets subtracted. A project's NPV nets out what you pay to start it, while this page subtracts net debt to get from the whole business down to the shareholders' share of it. Enter a market price and you get the NPV framing directly, because value per share minus that price is what you would be gaining, or overpaying, at today's quote.
Why terminal value dominates the answer
Look at the split bar. On the steady preset, 78% of enterprise value sits in terminal value, the single number standing in for every year after the forecast ends. On the capital intensive preset it reaches 85%, because a low discount rate does less to shrink distant cash.
Now compare the high growth preset, where terminal value is only 65%. The business is no easier to value. The forecast simply runs ten years instead of five, so more of the value gets modeled explicitly rather than compressed into one assumption. Extending the forecast window is the main lever analysts have on this, and Try 4 below shows it directly.
This is not a flaw in the model, it is what valuing a going concern means. But it has a consequence worth sitting with: building a careful five year forecast and then attaching a terminal value from two typed assumptions means the careful part is the minority of the answer.
Why WACC is built from its parts rather than chosen
WACC is what the people funding the business require, blended by how much each provides. Equity holders are paid last, so they demand more, and the capital asset pricing model estimates that as the risk-free rate plus a premium scaled by beta.
Debt is cheaper twice over. Lenders are paid first, so they accept less, and interest is tax deductible, which is why the model multiplies the cost of debt by one minus the tax rate. Load the capital intensive preset to see this: a utility carries far more debt than a software company and gets a materially lower discount rate for it.
The reason to build the rate from parts rather than type in a number is that every part is contestable. Beta depends on the window you measure it over. The equity risk premium is estimated rather than observed. Two careful analysts can defend rates a full point apart, and a point of WACC is worth a great deal.
The formula that breaks, and why the break is honest
Set terminal growth above the discount rate and the model refuses. The perpetuity formula divides by WACC minus growth, so as growth closes on the discount rate the value runs to infinity, and past it the value goes negative.
That looks like a technicality and it is actually the model enforcing something real. A company growing faster than the economy forever eventually becomes the entire economy. This is why terminal growth belongs at or below long run nominal GDP growth, usually somewhere near two to three percent, no matter how good the business is today.
Sensitivity analysis is the real output
The headline figure at the top is the least useful thing on this page. The sensitivity grid is what an analyst actually presents, and it is worth understanding why.
Move the discount rate by a point and a half in each direction, and terminal growth by one point, and value per share often swings by half its own size. Every cell in that grid is defensible. Quoting the middle one to the cent implies a precision the model cannot support.
The useful way to use a DCF is backwards. Rather than asking what the company is worth, ask what would have to be true for today's price to make sense, then judge whether those assumptions sound reasonable. That reframing turns an unfalsifiable number into a question you can actually argue about.
What this model leaves out
A great deal, and some of it matters. There is no balance sheet, so nothing checks that the capital spending you assume is consistent with the depreciation, or that the company can fund the growth you gave it. Real models link three statements precisely so those errors surface.
Capital structure is fixed, when in practice it drifts and beta should be relevered to match. There are no operating leases, pensions, minority interests, or dilution from stock compensation, all of which move the bridge from enterprise value to equity value. Mid-year discounting is ignored, which understates value by roughly half a year of return. And forecasting a single path says nothing about the distribution around it.
None of that rescues a bad forecast. The largest source of error in almost every DCF is the revenue and margin assumptions, not the mechanics.
Things to try
Each takes a few seconds and lands something the formula alone will not.
Break the perpetuity
On any preset, raise terminal growth until it passes WACC. Watch the model refuse rather than print a negative number, and read why.
Find the point of WACC
Change beta from 1.0 to 1.3 and nothing else. Compare value per share before and after. That is the cost of one contestable input.
Make the two methods disagree
Note the value under perpetuity growth, then switch to exit multiple. Then find the multiple that reproduces the perpetuity answer, which is what a sanity check looks like.
Shrink the terminal share
On the high growth preset, extend the forecast from ten years to fifteen. Watch the amber block shrink as more of the value gets modeled explicitly.
Value a shrinking company
Load the melting ice cube. Revenue falls every year and terminal growth is negative, and it is still worth real money. Declining is not the same as worthless.
Reverse the question
Leave the market price set and adjust growth and margin until value matches it. You have just worked out what the market is assuming.
Unfamiliar term? The course glossary defines the fifty this site actually uses, in the sense this site uses them.
Common questions about DCF valuation
What is a DCF and what is it actually telling you?
A discounted cash flow model says a business is worth the cash it will produce, with future cash counted for less than cash today. You forecast free cash flow for several years, add a terminal value covering everything after, discount it all at the cost of capital, and subtract net debt to reach equity value. The important caveat is that it is a statement about your assumptions rather than a fact about the company, which is why two analysts with the same model can reach values a long way apart.
Why is terminal value such a large share of the answer?
Because a going concern does not stop when your forecast does. Even after discounting, the cash from year eleven onward usually outweighs the first ten years, which is why terminal value commonly lands between 60 and 80 percent of enterprise value. The uncomfortable consequence is that most of the answer rests on the part you did not model in detail, which is the main argument for longer explicit forecasts.
What is WACC and why is debt cheaper than equity?
WACC is the blended return that lenders and shareholders together require, weighted by how much of the capital each provides. Debt is cheaper for two separate reasons: lenders are paid before shareholders so they take less risk and demand less return, and interest payments are tax deductible, which the model captures by multiplying the cost of debt by one minus the tax rate. That second effect is why heavily indebted, stable businesses like utilities carry noticeably lower discount rates.
Why can terminal growth not exceed the discount rate?
The perpetuity formula divides by the discount rate minus the growth rate, so as growth approaches the discount rate the value heads toward infinity, and above it the value turns negative. That is arithmetic, but it reflects something true: a company growing faster than the economy forever would eventually become the entire economy. Terminal growth belongs at or below long run nominal GDP growth regardless of how strong the business looks today.
Why do analysts show a sensitivity grid instead of a single number?
Because one figure implies a precision the model does not have. Moving the discount rate by a point and terminal growth by half a point can change value per share by half its own size, and every one of those inputs is defensible. Presenting the grid makes the range explicit and shifts the discussion from what is this worth to which assumptions would have to hold, which is a question people can actually argue about productively.
What is the difference between enterprise value and equity value?
Enterprise value is what the operating business is worth to everyone who funded it, lenders included. It is what you get by discounting unlevered free cash flow, which is measured before interest. Equity value is what remains for shareholders once net debt is subtracted, and dividing that by the share count gives value per share. Mixing the two up, for example comparing enterprise value to a share price, is one of the most common errors in valuation work.
Should I use perpetuity growth or an exit multiple?
Use both. Perpetuity growth is grounded in theory but extremely sensitive to a rate nobody can observe. An exit multiple is grounded in what comparable businesses actually trade for but imports whatever the market currently believes, including its mistakes. Running both and checking that they land in a similar range is a standard sanity check, and a large gap between them is a signal that one set of assumptions needs revisiting.
Is a DCF the same thing as net present value?
Nearly. Both discount future cash flows at a rate that reflects their risk and then add them up, so the machinery is identical. Net present value goes one step further and subtracts what you pay: a project needing $10m up front against $12m of discounted inflows has an NPV of $2m, and a positive NPV is the case for doing it. This page stops at present value to produce enterprise value, then subtracts net debt rather than a purchase price, because it is valuing an entire business instead of accepting or rejecting one project. Enter a market price and you get the NPV version of the question, since value per share minus price is what you would gain or overpay at today's quote.
About this model: a single-scenario unlevered DCF with a fixed capital structure, no linked balance sheet or cash flow statement, no mid-year discounting, and none of the adjustments real work requires for leases, pensions, minority interests or share-based compensation. Use it to understand how the pieces fit together and how sensitive the result is, not to value a specific company for a real decision.