The paper opens by putting its whole case in seven sentences, on the view that an argument you have to read twelve pages to find is not an argument. They are summarised here.

• You know what you bought. You do not know what you used. Purchase data is complete and always recoverable. Consumption data exists for one instant, at the store, and is gone if nothing records it.
• The cost of that gap is real and continuous, and it never appears as itself. It shows up later as a budget overrun, a slipped schedule, a write-off or an audit finding — in four different departments, none of which can see the cause.
• It cannot be measured backwards. You can start measuring today. You cannot decide today to have measured last year, which is why every month of delay is a month permanently missing from the evidence.
• A spreadsheet cannot do this job — not because people are careless, but because it is written after the event, by someone who was not there, and it cannot stop anything leaving the store.
• Nor should one person carry it. A storeperson is a gate that is open eight hours a day, and the system lives in their head until the day they leave. That is not a criticism of them. It is an unfair position to put them in.
• You are not paying to collect information — you are paying for the control. The record is what the control produces on its way past. You would want the locked door even if the record were thrown away.
• And you can test the whole case with one division. The fee against the controlled spend gives the reduction it has to deliver to pay for itself. Decide whether that number is plausible. Nothing else in the argument needs to be believed.

Each of these accumulates every day, in small increments, against no line item. None of them is anybody's fault and none of them triggers an alarm. Every one is eventually recognised under a different name, which is precisely why they are never fixed at source. The paper sets out all six with how each accrues and how each finally surfaces; the two below are the ones people recognise first.

The second half is written for the person who has to approve it. It works the arithmetic in full with nothing hidden, sets out what it costs beyond the invoice, states plainly what is actually being approved, and lists six questions worth putting in writing before anyone signs. It closes on something worth saying out loud: the case is usually sound, and the shape of the ask is usually wrong — buy a term rather than a commitment, get the category right, and work backwards from the budget date.

Figures, tables and the worked example are in the document, not on this page. No rates are published on this site; every figure is sized from the three drivers and confirmed at a short scoping review.