Once I have established what decision a model has to inform, the next question is which handful of numbers will decide the answer. Get that wrong and the effort goes into places that cannot change the outcome.

‍

A mining project has dozens of moving parts. Commodity price, throughput, recovery, cost structure and capital intensity are the typical candidates. Which of them actually drives this project is a separate issue.

‍

How to Find the Ones That Matter

‍

Ranking variables by their effect on NPV is the standard first step, and it is only half the answer. A variable with significant leverage but a narrow credible range is not a driver — it will land close to where you expect. A variable with less leverage but a wide credible range can matter far more.

‍

The two multiply:

• how far the variable can credibly move, given what is actually known about it
• how much the outcome moves when it does

‍

The company tax rate has real leverage and almost no range; it is legislated. Plant utilisation carries a far wider range, and it is seldom tested. The second is the driver.

‍

When everything looks important, nothing is.

‍

‍

What You Can Move and What You Carry

‍

A list ranked on leverage and range raises a second question: which of these can you do anything about?

‍

Throughput, recovery, capital cost and operating performance are yours to influence. You can test, quote or design around them. Commodity price, exchange rates, cost inflation and tax rates are not. You can hedge a few, contract around more, and fund the project to carry the rest, but none of that moves the thing itself.

‍

Two drivers can carry the same leverage and the same range and still call for different responses. One is a work programme. The other is a position to structure around.

‍

Turning Sensitivity into a Decision

‍

A base case produces a number. Sensitivity tells you how much of the credible range still clears the hurdle.

‍

That is the decision. A project that works only at base case is a different proposition from one that holds across most of what could reasonably happen.

‍

Where the shortfall sits in drivers you can move, there is a work programme: more testwork, further drilling, a firm quotation. Each costs money and buys a narrower range, and the ranking tells you which is worth funding. Where it sits in what you carry, there is no programme to buy.

‍

You structure around the exposure, or you do not proceed.