What it will cost

Somebody has to put a number in next year’s budget, and the figure you have been handed is either a price per person per month or a range with the word usage in front of it. Neither can be defended in the room you will defend it in, because neither shows you the quantities that produced it. These twelve pages are the quantities, in the order that makes each one legible.

The narrative account comes first because it is the only page here that shows the whole meeting at once: what your finance director will ask, and why the answer is a set of multiplications rather than a rate. Read it and the eleven pages after it stop being a reading list and become the parts of one conversation.

The next four take one multiplication each and slow it down. How many rounds a piece of work takes. How many advisers each round consults. What the machine must be handed before it can answer at all, which moves the bill most and gets a single line in a proposal. And the one line of text that makes that reading cost many times more. Each is the reason the next is worth reading: a round is expensive because of what happens inside it, what happens inside it is mostly reading, and reading has one failure mode that does the damage.

A ceiling on rounds bounds how long a piece of work may run, and it does not bound what that work costs. A council caught a page implying otherwise, and the correction sits inside the paper now.

The three after that are dials somebody set before you saw a screen: where material found mid-conversation may go, how hard the thing may think and about what, and which model takes which call. They group together because each is a choice with a price attached, each was made by a person, and none of them appears on a commercial page.

Then two illustrations of a bill nobody could reproduce. Each takes an invented company and says so in its own caption, and each describes a system that works correctly, passes every demonstration, looks identical on screen, and bills in a way nobody can explain. They come late rather than early because a failure only teaches once you can see which multiplication it landed in, and both of these land in the same one.

The last two are short, and they are the asks. One is the artefact that settles the argument. The other establishes how much of what you are buying is the reasoning at all, which is what makes one quote comparable to another.

If the whole path is more than you have, read the opening account and the two case notes. That is about twenty-two minutes, and it leaves you able to ask for one named piece of work priced into what was read and what was written, with the rounds counted.

What this path cannot do is tell you what your bill will be. Nobody can, without your volumes, your document sizes and the length of a typical piece of work in your company, and a page that produced a figure from an invented company would be selling you comfort rather than a method. Modelled scenarios at a range of company sizes are a separate cluster, built so the arithmetic moves when the inputs do, labelled modelled wherever they appear, and still no substitute for your own numbers. What these twelve pages give you is the ability to read a quote and say which multiplication in it has a ceiling you can read.