The discount rate is the annual return you require in exchange for putting money into an asset and waiting. In a valuation it does the work of converting future cash into present cash: divide by one plus the rate for every year of waiting, and a distant payment shrinks to what it is worth to you now.
It carries two ideas at once. The first is opportunity cost — money in this business is money not earning a safe return elsewhere. The second is risk — the less certain the cash, the more return you should demand for accepting it. A higher discount rate is therefore both a statement about alternatives and a statement about doubt, and it always lowers the valuation.
Where the rate comes from depends on whose cash is being discounted. For cash to shareholders it is the cost of equity, usually built with CAPM:
For cash to the whole business, debt and equity together, it is the weighted average cost of capital. Many investors skip the machinery and simply use the return they personally require — which is a defensible choice, provided it is applied consistently.
A payment of $1,000 five years away, discounted at three different rates:
The same future dollar is worth a third less at 14% than at 6%, and the gap widens the further out the cash sits. That is why the discount rate is usually the single most powerful input in a long-horizon valuation — it compounds against the forecast.
Pick a rate and keep it. Comparing two companies at two different discount rates compares your moods rather than the businesses; if one genuinely deserves a higher rate, say why in one sentence before you change it.
Match the rate to the cash flow. Discount cash available to shareholders at a cost of equity and cash available to the whole business at a weighted average cost of capital — mixing them double-counts or ignores the effect of debt.
In the app, Simple DCF and the Dividend Discount Model take the rate as a single editable figure, and Advanced DCF builds one from a risk-free rate, a beta, a market risk premium and a cost of debt.
Its influence grows with the horizon. On a two-year view the rate barely moves the answer; on a perpetuity it is half the formula, and the gap between the discount rate and the assumed growth rate is the valuation. Any model whose value sits mostly in a terminal value is really a model about this one number.
There is no correct value and no way to observe one. CAPM lends it a scientific air it has not earned: beta is measured over an arbitrary window, the market risk premium is a convention rather than a measurement, and small changes to either move the output more than most changes to the business would. Using one rate for every year also assumes risk is constant over the whole forecast, which for a young company it plainly is not.