The core idea
A dollar next year is worth less than a dollar today, because today’s dollar can be invested and because the future one might not arrive. A discounted cash flow puts a number on that: project the cash a business will generate, shrink each year’s figure by how far away and how uncertain it is, and add up what remains.
The cash being projected is free cash flow, not earnings. Earnings are an accounting result and can be produced without cash arriving; free cash flow is what is actually left after the business pays to sustain itself.
The four steps
One: project free cash flow. Take the most recent year and grow it at a rate you can defend, usually for five to ten years. Beyond a decade, forecasting is guesswork wearing a spreadsheet.
Two: pick a discount rate. Usually weighted average cost of capital, which blends what shareholders and lenders each require. It carries both the time value of money and the risk that the forecast is wrong.
Three: value everything after the forecast. The terminal value assumes cash flow grows at a modest constant rate forever, and it is discounted back like any other year.
Four: convert to a per-share value. Add the forecast and terminal present values for enterprise value, subtract net debt to get equity value, then divide by shares outstanding.
Worked example. A company generating $100M of free cash flow, growing 8% for ten years, then 2.5% forever, discounted at 10%, with 50M shares and no net debt. Enterprise value comes to $2.04B, or $40.86 a share. Run it yourself in the DCF calculator.
The terminal value problem
In that example, 55.7% of the total value comes from the terminal value, the part representing every year after year ten. That proportion is typical, and it is often higher.
It is worth sitting with what that means. More than half of the answer rests on a single assumption about what happens in a decade and beyond, which is the period you know least about. The ten years of careful projection you just built contribute less than the one number you picked at the end.
How fragile it is
Small changes to inputs move the answer a great deal. Holding the same company constant and changing one assumption at a time:
| Change | Fair value per share | Difference |
|---|---|---|
| Base case: 8% growth, 10% discount | $40.86 | — |
| Growth cut to 5% | $32.79 | −20% |
| Discount rate raised to 12% | $31.46 | −23% |
A three-point change in growth removes a fifth of the value. Two points on the discount rate removes nearly a quarter. Neither adjustment is remotely unreasonable, which is why two analysts can build honest models of the same company and land 40% apart.
That sensitivity is also why a DCF is so easy to abuse. It takes very little tuning to reach whatever answer you wanted, and the output looks authoritative regardless.
Using it honestly
Use ranges, not points. Run a low, base, and high case and treat the spread as the actual output. A single number implies a precision the method cannot support.
Then apply a margin of safety. The cushion exists precisely because the assumptions above will be partly wrong, and it is what converts a fragile estimate into a usable decision.
Best of all, invert the question. A reverse DCF holds the market price fixed and solves for the growth rate it implies. You then judge one number against the business instead of defending five, and there is nothing left to tune.