Day three of the close. Aisha's supplier reconciliation shows a difference of exactly one hundred and forty-eight thousand pounds, and nobody knows why. Three people test one explanation, drop it, find the real one, and then realise it may not be the only one.
Nadia, sorry, have you got a minute? You said to interrupt you.
At the kickoff Nadia said, "You interrupt me." Aisha reminds her of it in the same sentence as she interrupts. This makes interrupting easier for a junior person: it is not rude, it is the agreed rule.
I did. Rotterdam can wait. What have you got?
Ferrand Metals. Their statement says we owe them four hundred and twelve thousand, three hundred. Our ledger says five hundred and sixty thousand, three hundred.
Aisha starts with the supplier and the two numbers. No introduction, no apology, no opinion yet. In a working session, the facts come first and the interpretation comes after.
So we think we owe them more than they think we do.
Nadia turns two numbers into one plain sentence. With a difference, the direction matters as much as the size: which side is higher tells you where to look.
By a hundred and forty-eight thousand. Exactly.
Exactly is interesting. A round number usually means one document, not lots of small ones. What's your guess?
Nadia asked at the kickoff for a flag with a guess. Now she asks for it. A manager who asks "what do you think it is?" is not testing you; she wants a starting point.
My best guess is timing. Something in transit at the year-end. But I'm not sure, and I haven't changed anything.
This phrase gives an opinion and marks it as uncertain in the same breath. Aisha adds the two facts her manager most needs: how sure she is, and that nothing has been touched.
Good. Don't. Let's get Matteo. Ferrand is one of his rebate suppliers, and I'd rather have him here before we start guessing.
A hundred and forty-eight on Ferrand? I can tell you what it is. Somebody posted an invoice twice. It happens every January.
Matteo answers before anyone has looked at the data. Every team has someone like this, and they are often right. The skill in the next line is how to test the answer without rejecting it.
Maybe. Let's not assume. If it's a duplicate, what would we see?
"If it's X, what would we see?" changes an opinion into something you can check. Nadia does not say Matteo is wrong. She asks what the evidence would look like if he were right.
Two invoices with the same amount and the same number, maybe a few days apart.
I can check that now. I'll sort all the December invoices by amount.
And while she does that, timing. If something was in transit, which way would it go?
If we'd paid them and they hadn't recorded it yet, their balance would be higher than ours. Theirs is lower. So it's not a payment in transit.
Matteo walks through the logic step by step: if this, then that; but the opposite is true; so not this. Saying your reasoning aloud lets others follow it and catch a mistake before it costs time.
Right. So if it's timing, it has to be something on their side that we don't have yet.
OK. No duplicates. The biggest December invoice is eighty-one thousand, and every invoice number appears once.
Hm. Then I was wrong. Not a duplicate.
"Then I was wrong" — four words, no excuse, no defence. Matteo lets his theory go as soon as the check fails. Finance teams trust people who can do this quickly.
That's fine. That's what checking is for. One theory gone, and we know more than we did five minutes ago.
Nadia reframes the failed check as progress. In an investigation, ruling something out is a result. Saying so keeps people willing to offer their next idea.
Aisha, go back to their statement. Not the balance, the lines. Is there anything on their side that we don't recognise?
Page one is all invoices, and the one payment in December. Page two. Oh. There's a credit note. Twentieth of December. One hundred and forty-eight thousand.
Aisha describes what she sees as she goes: page one, page two, then the find. Narrating a search keeps everyone in the room with you, and it is natural to do in a working session.
What does it say?
"Volume rebate, tier two, full year."
Tier two. So we did cross the next tier with Ferrand.
That's one of your two unknowns from the kickoff.
It is. I estimated Ferrand at tier one. Ninety-six thousand.
Matteo states his own estimate plainly, even though the new document shows it was low. He does not explain or defend it yet. He gives the figure so the team can compare.
What did you assume, and what does tier two need?
Tier two starts at four million pounds of purchases in the year. My December estimate put us at three point nine.
Our ledger says four point one for the year. I can see it in the supplier report.
Then my December guess was two hundred thousand too low. That's the whole story of the tier, right there.
Nadia asks two neutral questions: what did you assume, and what is the rule? The answer explains the gap without anyone having to defend themselves. Matteo then names his own error in one sentence.
So, before we go any further: is this credit note in our system at all?
A useful phrase to stop a discussion from running ahead of the facts. Nadia checks the most basic question first: does the document exist on our side?
Let me search by the credit note number. Yes. It's here. It was posted on the second of January.
The second of January. So it's dated December, it belongs to December, and it's sitting in January.
Nadia states the whole problem in one rhythm: dated, belongs, sitting. Short parallel clauses are easy to follow when you are explaining something technical to people under time pressure.
Why would someone do that?
I don't think anyone did it on purpose. My bet is that payables post documents by the date they receive them, not the date on them. It arrived after Christmas, so it went into the new year.
Aisha's question could lead to blame. Nadia moves it straight to the process: how documents are dated. "Payables" is short for the accounts payable team, who record supplier invoices and credit notes.
So if we move it back into December, the difference on the reconciliation disappears.
It does. But then we have a different problem. Your accrual.
Nadia agrees first, then adds the complication. Agreeing before you add a problem keeps the other person with you; starting with "no" makes them defend their point.
Ah. Yes. I've accrued ninety-six for Ferrand. If the credit note goes into December as well, we count the same rebate twice.
One place only. The credit note replaces your estimate for Ferrand.
So the real change isn't a hundred and forty-eight. It's the difference between my ninety-six and their hundred and forty-eight. Fifty-two thousand.
The reconciliation showed a difference of one hundred and forty-eight thousand, but the effect on the accounts is fifty-two. Matteo separates the two numbers. Graham's first question will be "how big is it?", and this is the honest answer.
So for my reconciliation, what do I write?
The difference has a name now. Credit note, its number, its date, posted in January, to be moved to December. That's a reconciled item, even before anyone moves it.
And do I attach anything?
Page two of their statement, and the credit note itself. If Alan asks about Ferrand next week, I want the answer to be one click away.
At the kickoff, Nadia said a reconciliation is done when every difference has a name. Here she applies it: Aisha can close her reconciliation as soon as the item is identified, even though the correction comes later. Alan is the external auditor who arrives after the close.
But that's good news, isn't it? It means our costs were lower.
It's good news for the profit. It's not good news for the close. A surprise is a surprise in either direction.
Aisha's point is fair: lower costs mean higher profit. Nadia's answer is the finance view: an unexplained change is a control problem, whether it helps the numbers or hurts them.
And I still have one more supplier where I don't know the tier.
That's not what worries me most, though. If payables post by the date they receive things, why would Ferrand be the only one?
Nadia steps back from one supplier to the process behind it. If the cause is how documents are dated, the same error could be in other places. This question is what turns a reconciliation item into something the director must hear.
I've finished fourteen of the twenty statements. I wasn't looking for credit notes on the others. I was only checking the balances.
And the balances agreed?
Thirteen of them did. One has a small difference, about two thousand. I was going to bring it to you this afternoon.
Bring it now, with the rest. A small difference nobody understands…
Is the one that grows. I remember.
Nadia starts her own rule from the kickoff and leaves it open. Aisha finishes it. This is a friendly way to show a rule has been learned, and it only works with a phrase the team already shares.
OK. Here's where we are. We know what the Ferrand difference is. We don't know yet whether it's the only one. So, three things.
Before giving out actions, Nadia summarises what is known and what is not. One sentence for each. The actions that follow then make sense to everyone.
Aisha, go back through the fourteen statements and look for any credit note dated December. For each one, write down whether it's in our system, and in which month.
And the six I haven't done yet?
Same check, as part of the reconciliation. If a supplier hasn't sent a statement, put it on the tracker as blocked.
And the two thousand? Do I treat it the same way, even though it's small?
Exactly the same way. Supplier, amount, your guess. If it turns out to be boring, that's the best possible answer.
In a close, an explanation that turns out to be dull is a relief, not a waste of time. Nadia keeps the same standard for a two-thousand-pound item as for a large one, because the size is only known once the item is explained.
And me?
Recalculate the rebate accrual with Ferrand at tier two, but don't post the true-up yet. I don't want two journals turning into four.
Nadia gives Matteo a task and a limit in the same sentence. The limit protects the audit trail: until the full picture is known, every new journal would need correcting later.
Fair. I'll have the number by one o'clock, with the basis.
"By one o'clock, with the basis." Matteo answers a task with a time and a standard. It tells Nadia exactly when she can plan around his number.
And the other supplier, the one where I don't know the tier. Should I wait for their statement?
No. Call them today and ask one question: have you issued a tier credit for last year, and if so, for how much?
One question. I like that.
Waiting for a document is passive; a phone call with one precise question is active. Nadia gives Matteo the exact question to ask, so the answer comes back in a form the team can use.
And I'll ask payables for every credit note they posted in the first week of January with a December date. That tells us the size of this, not just the shape.
Do we tell Graham now, or when we know the size?
Now. Well, at two. That's what he asked for. We suspect something, we don't know how big it is, and it's day three, not day five.
Matteo offers the easier option: wait until we are sure. Nadia applies Graham's rule from the kickoff: tell him the day you suspect it. "It's day three, not day five" is the whole argument.
He won't like it.
He'll like it more on day three than on day five.
Who's telling him?
I am. You're coming with me, with your number.
Should I come?
Not this time. But it was your flag, and I'll tell him that. Your guess was half right. The flag was completely right. That's how it's supposed to work.
Nadia separates two things: Aisha's guess, which was half right (it was a timing problem, but nothing was in transit), and her action, which was completely right. Saying both, in that order, gives honest feedback and real credit at the same time.
Thanks. I'll start with the fourteen.