Why can’t treasury explain the cash position on the spot?
Because in most groups the position is assembled rather than produced. Balances come from the banks, postings come from the ledger, payment status comes from a third place, and the movements between entities come from nowhere in particular. Nobody built that number; somebody stitched it.
Stitching works until the question changes. A total can be produced by hand. An explanation cannot, because it needs every input to still be identifiable after the addition: which statement line, which entry, which run, which entity. The moment those identities are lost in a paste, the only way back is to do the whole thing again.
That is why the honest answer to “why did cash move” so often arrives the next day. Not because the team is slow, but because the question is asking for something the process never kept.
What is the difference between seeing cash and controlling it?
Seeing is a balance. Controlling is being able to say what that balance is made of and what will move it before Friday. A dashboard delivers the first and gets mistaken for the second.
The test is a set of questions rather than a screen. What changed since yesterday, and how much of it was operating, financing or one-off? Which entities consume liquidity every month rather than occasionally? Which figures has a bank confirmed, and which are still the ledger’s opinion? A team that can separate those is controlling the position. A team that can only total it is reporting one.
The distinction shows up in the week it gets expensive. Anybody can explain a quiet month.
Where does the explanation actually break?
At the joins. Each source describes the same day in its own vocabulary, and the position is where those vocabularies are forced together.
A bank sends one line for a settlement the ledger booked as forty. The ledger records a payment run on the day it was approved; the bank reports it on the day it left. An internal transfer is one movement to the group and two entries to the two entities, in two currencies, on two dates. None of that is an error. It is four correct systems describing one event differently.
So the work is not addition. It is deciding, line by line, which entry each movement is the same event as, and keeping that decision where somebody can look at it later. A process that decides in a spreadsheet keeps the total and throws away the decision, which is exactly the part the CFO is asking about.
What does an explainable position look like?
It looks like a figure you can open. Every total resolves to the movements underneath it, every movement to the statement line and the journal entry it was matched to, and every match to the evidence that made it the best explanation, together with the alternatives that scored lower.
That last part is the one usually missing. Matching is not a lookup, it is a judgment between competing explanations, and a system that keeps only the winner leaves the reviewer holding an assertion. Keeping the rejected explanations with their scores is what turns “these two agree” into something a second person can audit.
It also changes what the morning is for. When the position explains itself, nobody spends the day defending yesterday’s number, which is the quiet cost that never appears on an invoice.