Finance.

The framework

1. 01

   Only the money the decision itself moves.

   A net present value runs on money the decision itself causes to move: what leaves for the asset and for the working capital it ties up, what comes back as incremental sales or as a cost that stops, and the tax paid on the difference. Depreciation isn't cash. Head-office overhead that carries on whether you do this or not isn't cash either. Neither is what you've already spent getting to the decision, which is gone down every branch of it. What sinks most of these models is an accrual discounted as though it were money in the account.

2. 02

   Discount each year at the rate you already set.

   Divide each year's cash by one plus your rate, raised to the number of years away. At a twelve per cent hurdle, a dollar arriving in year three is worth about seventy-one cents this morning and a dollar in year five about fifty-seven. Add the discounted years, subtract what you pay now, and what remains is the value created in today's money over and above clearing the bar. Use the rung you assigned this project on Day 156, before the model existed. A rate reverse-engineered until the answer turns positive isn't a hurdle, it's a decision already taken looking for paperwork.

3. 03

   Decide on the dollars; treat the percentage as a witness.

   The internal rate of return is the rate at which net present value falls to zero, and people reach for it because a percentage feels comparable across everything. It quietly misleads across different sizes and different lives: sixty per cent on forty thousand and eighteen per cent on nine hundred thousand can't be ranked by percentage when you hold nine hundred thousand and get one shot with it. Payback carries the opposite bias, preferring whatever returns fastest and going blind the day the money is back. Decide on present value in dollars; let the other two corroborate or object.

4. 04

   Every case is hanging on one assumption. Which is yours?

   Most of the answer lives in a handful of inputs, and usually in the later years where the estimate is thinnest. So don't present a number, present the point at which it breaks. Walk each major input — the volume, the price, the months to ramp, the rate itself — until present value crosses zero, and write that break-even value beside it. Then ask the only real question: is that something this team has ever actually done? A five-decimal answer resting on a growth rate somebody chose in a meeting is precision doing an impression of confidence.