Nine years signing off corporate inventories at an assurance firm, before deciding the working ought to be published with the number.
Every engagement starts in the same place, and it is not the spreadsheet. It is a conversation about where the boundary falls — which sites count, which joint ventures count, and which of the things you own are not really yours to report. That conversation is written down before anything is counted, because a boundary agreed halfway through a measurement is the single most expensive thing that can happen to a figure.
What follows is deliberately unglamorous. Meter exports are read against purchase ledgers. Supplier responses are chased, and where they do not arrive, the gap is recorded as a gap rather than filled with an average. By the end there is a number, and beside it a file that explains how the number was reached, including the parts of it we are least happy with.
Clients are often surprised by the shape of the effort. The counting itself is quick. What takes the weeks is establishing that the inputs are what they claim to be, and it is also the part that decides whether the figure survives an assurer reading it closely a year later.
At the end of an engagement the model, the raw data and the assumption log are handed over in a format that opens without us. This is not generosity; it is the only arrangement under which a figure can honestly be described as the client's own. A number you cannot reproduce is a number you are renting.
They opened with the three numbers in our last report they could not stand behind. An uncomfortable hour, and it saved us a correction.
Not every piece of work should be bought. Where an existing inventory already holds up, we say so and quote for the smaller job of assuring it. Where a client wants a figure faster than the data can honestly produce one, we explain what the shortcut would cost later, and let them decide with that in front of them.
