Six years as an external assurer before joining, which is why he knows in advance which figure will be picked out at random.
An assurance engagement has a rhythm. The assurer asks for the boundary note, then for one figure chosen more or less at random, then for the file that figure came from. If the file takes a week to assemble, the engagement has already gone badly, and the next question will be sharper than the last.
So the file is built in advance. Every number in the inventory traces back to the export, invoice or meter reading it rests on, and anything modelled is labelled as modelled before anyone else has to point it out.
Not a summary. The working, indexed the way somebody looking for a specific number would look for it, which is rarely the order the inventory was built in. An assurer arrives with one line in mind, and should reach the evidence behind it in a minute rather than after a morning of searching.
Questions arrive faster than a finance team can answer them while also doing its own job, and an answer improvised under pressure is how a correction gets made the following year. Somebody who has run the numbers should be there to say plainly which ones are solid and which ones are estimates.
I spent six years asking these questions. The uncomfortable ones are always the same eleven, and nine of them can be answered before the meeting starts.
It does not mean the figures were checked lightly and found correct. It means a narrower set of procedures was performed. Where a client wants that distinction softened in the write-up, we decline, because the distinction is the only thing the opinion is actually saying.
