Sanity-Checking a Finished DCF

A DCF can pass every formula check, the discounting is correct, the terminal value formula is applied properly, the model ties out to the penny, and still imply that a mid-size regional retailer will be worth more than the entire US economy in forty years. The model isn't broken; the assumptions feeding it are. Checking a DCF's internal arithmetic is necessary but nowhere near sufficient, the numbers it spits out need to be checked against what's actually plausible in the real world.

The checks that catch the most damage

Terminal value as a share of total value. If the terminal value makes up 85–95% of a DCF's total value, nearly the entire valuation rests on assumptions about a period more than a decade out, which should be flagged and stress-tested even if the near-term forecast is well-researched.

Perpetual growth rate versus the economy. A company's cash flows cannot grow faster than the overall economy forever, a terminal growth rate above roughly the long-run nominal GDP growth rate (commonly 2–4% for developed economies) implies the company eventually becomes the entire economy, which is not a plausible steady state.

Implied return on capital in the terminal year. If the terminal-year numbers imply a return on invested capital far above what any company in the industry has sustained historically, the growth and margin assumptions feeding the model deserve a second look.

Cross-check against multiples. If the DCF's implied valuation is wildly out of line with what comparable companies actually trade at, that's not automatically a sign the market is wrong, it's a prompt to check whether the DCF's assumptions (growth, margin, discount rate) are actually defensible, or whether an error has crept in.

Worked example

A DCF for a consumer goods company produces an enterprise value of $8.0 billion, of which the terminal value contributes $7.4 billion.

1. Terminal value share: $7.4/8.0 = 92.5\%$, a flag that nearly the entire valuation rests on the terminal-year assumptions.
2. Checking the terminal assumptions: the model uses a 4.5% perpetual growth rate, above the roughly 2.5% long-run nominal GDP growth typically used as a ceiling, a second flag.
3. Checking implied ROIC in the terminal year: the model's terminal operating income and invested capital imply a 35% return on invested capital, well above this industry's historical range of 12–16%, a third flag, and now a pattern.
4. Cross-check against peers: comparable consumer goods companies trade at 12x EV/EBITDA; this DCF's $8.0 billion implies roughly 19x the company's current EBITDA, well outside the peer range, consistent with the three flags already raised.

Lowering the terminal growth rate to 2.5% and the implied terminal ROIC to a level consistent with peers brings the DCF down to roughly $5.1 billion, in line with the peer multiple range, and a far more defensible number than the original $8.0 billion.

What this means in practice

Sanity checks are cheapest to run, and most valuable, at the very end of a DCF build, once every number is in place, running them earlier means re-checking after every subsequent change, but skipping them entirely is how obviously wrong valuations survive into a final report. A DCF that fails several of these checks isn't necessarily useless, but every failed check should trace back to a specific assumption that gets revisited and explicitly justified, not silently left as-is.