Hold on to this
The number the model produces isn't the intrinsic value. It's one scenario. The job is to run three.
Instrument · Drive it
Try it yourself
DCF Sensitivity Matrix
| disc. ↓ / term. → | 1% | 2% | 3% | 4% |
|---|---|---|---|---|
| 8% | — | — | — | — |
| 9% | — | — | — | — |
| 10% | — | — | — | — |
| 11% | — | — | — | — |
| 12% | — | — | — | — |
| 13% | — | — | — | — |
Enter FCF per share and a 5-year growth estimate to populate the grid.
5-yr DCF + Gordon Growth terminal value · Columns: terminal growth rate · Rows: discount rate · Illustrative only.
The problem DCF is solving
A stock is a claim on future cash flows. The problem is that future cash flows aren't worth the same as present ones. A dollar promised five years from now is worth less than a dollar today — partly because of inflation, partly because a lot can go wrong before the payment arrives, and partly because if you had the dollar today you could invest it elsewhere. The further out a cash flow sits, the more you should discount it.
Discounted cash flow analysis is how you price that uncertainty. You estimate how much free cash flow a business will generate over a forecast window, apply a discount rate to convert each future year's cash into present-day dollars, add a number representing everything after the forecast window closes, and sum the pieces. The output is what those future dollars are worth today — and therefore what you should pay for a share of the business.
The discipline the method provides is that every assumption is explicit. A price-to-earnings ratio lets you anchor to whatever the market is paying and call it valuation. DCF does not. You have to name a discount rate, name a growth rate, and name a terminal assumption. Those numbers will be wrong. That is expected and fine. Being forced to commit them to a model is the point.
The four inputs that drive everything
Every DCF model runs on four numbers. The first is free cash flow per share — trailing twelve months, after the business has paid to maintain and grow its asset base. Use FCF, not earnings. Earnings are an accounting output; FCF is what ends up in the bank.
The second input is the near-term growth rate: how fast FCF will compound annually over your forecast window, typically five to ten years. Most analysts spend the most time here. They shouldn't. Near-term FCF projections drive a minority of the total output. A useful anchor for this input is the company's historical revenue or FCF CAGR — measure it across multiple periods with our free CAGR calculator before setting your growth assumption.
The third is the discount rate — typically WACC, the weighted average cost of capital. This is the rate you use to convert each future dollar into a present-day equivalent. At 10% WACC, a dollar received in five years is worth $0.62 today. At 8%, it's $0.68. That four-cent difference per dollar compounds across ten years of cash flows and a terminal value. Small WACC differences produce large output differences.
The fourth input is the terminal growth rate — how fast the business grows in perpetuity after the forecast window closes. This is the single assumption with the most leverage on the output, and it deserves the most scrutiny.
A DCF from scratch: one company, real numbers
Take a fictional mid-cap industrials company — Meridian Components — trading at $58 per share with trailing free cash flow of $4.00 per share. You project 8% annual FCF growth over five years, use a 10% discount rate, and assume 2.5% terminal growth. Here is every step, one year at a time (illustrative numbers):
- Year 1 FCF: $4.00 × 1.08 = $4.32 → discounted at 10%: $3.93
- Year 2 FCF: $4.32 × 1.08 = $4.67 → discounted: $3.86
- Year 3 FCF: $4.67 × 1.08 = $5.04 → discounted: $3.79
- Year 4 FCF: $5.04 × 1.08 = $5.44 → discounted: $3.72
- Year 5 FCF: $5.44 × 1.08 = $5.88 → discounted: $3.65
Five years of discounted cash flows sum to $18.95. Now the terminal value. Take year-5 FCF ($5.88), apply one year of terminal growth (× 1.025 = $6.03), and divide by the spread between discount rate and terminal growth rate: $6.03 ÷ (10% – 2.5%) = $6.03 ÷ 7.5% = $80.40. That is the terminal value as of year five. Discount it back five years to today: $80.40 ÷ (1.10)⁵ = $49.91.
Total intrinsic value per share: $18.95 + $49.91 = $68.86. At $58, the stock looks 19% cheap. Terminal value contributed $49.91 of that $68.86 — 73%. The five years of careful projection work contributed 27%.
If you want to replicate this in a spreadsheet, the steps are:
- Multiply FCF per share by (1 + growth rate)^year for each year in your window
- Divide each year's FCF by (1 + discount rate)^year to get its present value
- Sum the present values across all forecast years
- Multiply the final year's FCF by (1 + terminal growth rate) to get terminal FCF
- Divide terminal FCF by (discount rate – terminal growth rate) to get terminal value
- Discount terminal value back: divide by (1 + discount rate)^N where N is the forecast window
- Add the discounted terminal value to the sum from step three
That is the complete model. It takes about fifteen minutes to build in a spreadsheet from a company's latest earnings release. The challenge is not the arithmetic. It is the inputs.
Why the number you just calculated is probably wrong
Take Meridian's $68.86 estimate and move two inputs. Change the discount rate from 10% to 12% — reasonable for a company in a cyclical industry with higher financial leverage. Change the terminal growth rate from 2.5% to 1.5% — reasonable if the industrial sector faces structural pricing pressure. New output: $50.56. At $58, the stock is now 13% expensive.
Run it the other direction: 8% WACC and 3.5% terminal growth. Also defensible for a company with stable recurring revenues and a strong balance sheet. Output: $111. Same company. Same facts. Three inputs moved by a combined four percentage points. The implied value went from $51 to $111.
The terminal value formula is fragile by construction. TV = FCF ÷ (r – g). When r and g are close, the denominator is small, and small changes in either input produce large swings in the quotient. A discount rate of 10% and terminal growth of 2.5% gives a denominator of 7.5%. Move terminal growth to 3.5% and the denominator falls to 6.5% — the terminal value expands by 15%. Since terminal value is 73% of total intrinsic value in the Meridian example, that 15% terminal value expansion moves the headline number by about 11%. From one percentage point.
This is why the sensitivity matrix in the tool above is more useful than any single output. It maps where the green-to-red boundary falls across realistic input combinations. The width of that band is the real information the model produces.
Stress-test your assumptions
The matrix above maps what happens to implied intrinsic value as discount rates and terminal growth rates shift across their realistic ranges. Enter your own FCF per share, growth estimate, and stock price. The grid recalculates on every keystroke.
Two things to watch: first, the green-to-red boundary isn't a line — it shifts considerably with small input changes, and that shift is faster than most investors expect. Second, no single cell is the answer. The distribution of cells — how many are green, how far into the red the bear-case corner goes — is what you're actually reading. A stock where 20 of 24 cells are green under honest input ranges is a different investment from one where only 4 cells are green and they require aggressive assumptions to justify.
What DCF cannot do
DCF requires a business that will eventually generate free cash flow in quantities you can roughly project. That excludes most early-stage companies, pre-revenue biotech, and any business whose competitive position is so unstable that forecasting cash flow growth feels like speculation. For those businesses, revenue multiples or gross profit multiples anchor you to something observable. DCF can work as a sanity check on a future exit scenario — what does the company have to earn in year seven for the current price to make sense — but it shouldn't be the primary tool.
Even when DCF is appropriate, it ignores several things that determine whether the cash flows you projected will actually show up:
- Competitive dynamics. A 7% growth assumption made today can collapse if a well-funded competitor enters and prices aggressively. DCF takes your growth rate as a given and runs the math. The market prices competitive response in real time.
- Management capital allocation. Two companies with identical near-term FCF profiles will produce very different long-run outcomes depending on whether management reinvests that cash at 20% returns or destroys it through dilutive acquisitions. The model cannot see this.
- Structural disruption. The terminal value assumes a going concern that grows at a stable rate forever. When the long-run economics of an industry change — streaming disrupting cable, EVs displacing ICE vehicles at scale — the perpetuity assumption breaks before the math does.
- Cyclical starting points. Projecting from a peak FCF year, or from a trough, systematically distorts the base. DCF is most reliable when you normalize free cash flow through the cycle before building the model.
Comparable company analysis fills some of these gaps by anchoring to what similar businesses actually trade for. Asset-based valuation works better for capital-intensive businesses where the balance sheet drives value. DCF is a useful discipline for thinking explicitly about growth and risk — not a complete picture on its own.
A decision rule that actually holds up
Run three cases. The bear case should represent a realistic bad outcome — tighter margins, slower growth, a higher discount rate — not a catastrophe scenario you could dismiss as improbable. The base case should reflect your actual expectation, not the average of bull and bear. The bull case should require something to go right that is not already visible in the price.
Buy discipline: only take a position when the bear case still implies a margin of safety — typically 15–20% below the current price. If you need the base case to work out perfectly to break even, the position depends on precision the model cannot deliver. If the stock only pencils out under bull assumptions, name exactly what has to be true and decide whether you believe it independently of the fact that you want the stock to be cheap.
The model's job is not to produce a target. It is to reveal which assumptions are load-bearing. If the bear case requires a discount rate above 11% to flip expensive, you are betting that the business stays stable enough to earn a low cost of capital. If terminal growth needs to hold above 3%, you are betting the company can outgrow the broader economy indefinitely. Name those bets. Defend them specifically, or pass.
Questions worth asking
What discount rate should I use?
Most practitioners use WACC — the blended cost of debt and equity. A simpler starting point: 8–10% for stable, capital-light businesses with predictable cash flows; 12–15% for faster-growing or riskier ones. The exact number matters less than running the sensitivity matrix across a range, because a 2-point shift in your rate can move the output by 30% or more.
Why do two analysts produce DCF valuations that are 50% apart for the same stock?
Because every input is an opinion, not a fact. Growth rate, discount rate, terminal multiple — all of them are judgment calls, and small differences in those calls compound across a 10-year projection. This is also why comparing DCF outputs across sources is nearly useless without understanding the underlying assumptions. The number means nothing; the assumptions mean everything.
Is DCF useful for companies with no profits yet?
Barely. The method requires free cash flow to exist at some point, so you're forced to project when the company turns profitable and at what margin — both of which are highly speculative for pre-profit businesses. Revenue or gross profit multiples tend to anchor you to something real. DCF can work as a sanity check on an exit scenario, but it shouldn't be the primary tool.
What is terminal value and why does it dominate the output?
Terminal value captures all cash flows beyond your explicit forecast window — typically everything after year five or ten — compressed into a single number. For most companies it accounts for 60–80% of the total DCF result, which means most of your valuation rests on one assumption about perpetual growth. That's why a 1-point change in the terminal growth rate moves the final number so dramatically, and why stress-testing it isn't optional.
How is DCF different from just looking at the PE ratio?
A PE ratio is a relative shortcut: it tells you what the market is paying per dollar of earnings compared to peers. DCF tries to be absolute: it tells you what the business is worth based on its own cash flows, independent of what anyone else is currently paying. DCF is slower and more assumption-heavy, but it forces you to think explicitly about growth and risk rather than anchoring to whatever the crowd is pricing in.
Keep reading