The last day of audit fieldwork. Alan brings his draft finding on the credit notes to Graham and Nadia. Nobody argues that it happened. The negotiation is about the words: which ones are accurate, which ones aren't, and what the company promises to do about it.
Thank you both. Before the draft, last year. Nadia, you sent me the result on the twentieth.
I did. The same problem existed last year. Six credit notes dated December were recorded in January. Sixty-four thousand in total.
How did you check it?
Aisha did the first pass on last year's top twenty statements, line by line. I pulled every credit note over five thousand recorded in the first two weeks of January, and checked the journals.
Line by line, even where the balances agreed?
Line by line, whatever the balance says.
Good. I'll take a copy of her working papers for my file.
Alan asks how, not just what. Nadia describes who did which part and how far they went. An auditor trusts a result more when the method is clear, and "whatever the balance says" shows the lesson from the one-to-one has become the way the team works.
And because they were recorded in January, they landed in this year's accounts. So this year's costs were too low by sixty-four, at the same time as they were too high by a hundred and sixty-eight.
Graham states both directions in one sentence: too low by one amount, too high by another. When an error crosses two years, saying both effects out loud prevents anyone from adding them up the wrong way.
And the sixty-four is below materiality, so you've listed it rather than corrected it.
Yes. It's on the list of uncorrected differences, with last year's documents attached.
Three weeks ago, Matteo said it would probably cancel out. I told him I didn't want "probably" written anywhere. Now it's written down, with evidence, and it doesn't fully cancel out. That's the difference.
In the escalation, Graham refused an argument with no proof behind it. Here the proof has arrived, and it shows the effect was real but partial. The point he makes is not who was right, but that a number is now backed by documents.
Agreed. And my sample of twenty-five credit notes from January and February found nothing new. Every one is in the right month since your change.
That's good to hear.
Just so we're clear on the scale before I read it. From least to most serious: observation, deficiency, significant deficiency, material weakness.
And this isn't a material weakness.
No. Your own process found it and corrected it before the accounts were finished. That's the line between the two.
Before discussing one rating, Alan names the whole scale. Graham then checks the most important boundary out loud. Agreeing what the words mean before using them avoids a long argument later.
So, the draft. I'll read it, and then you tell me what you'd change. "Significant deficiency. Credit notes were recorded in the wrong period due to a failure of the cut-off control, resulting in a material misstatement."
Alan reads his exact wording and invites changes. In audit closing meetings, findings are often agreed word by word. The question is never only "do you agree?", but "what would you change, and why?"
Thank you. Let me say first what we're not going to argue about. It happened, it was above materiality, and it's a significant deficiency. We accept all three.
Graham starts by saying what he accepts. This removes the auditor's need to defend those points and makes the rest of the conversation about details, not principles. It also shows that the requests that follow are about accuracy, not about escaping the finding.
That's helpful. And for the record, the rating isn't about how you handled it. It's about what could have happened if nobody had read that statement.
But there are two phrases that I don't think are accurate. Nadia, the first one is yours.
"A failure of the cut-off control." There wasn't a cut-off control for credit notes. Payables recorded them by the date they received them, and nobody checked. Saying the control failed suggests that one existed.
That's an unusual request. You're asking me to make it sound worse.
I'm asking you to make it accurate. If someone reads this next year and looks for the control that failed, they won't find one.
Nadia asks for a wording that is less flattering to the company, because it is true. "I'm asking you to make it accurate" is a strong sentence in any negotiation: it moves the discussion away from who wins and towards what is correct.
It's less flattering. It's also true. Use it.
All right. "Due to the absence of a cut-off control for credit notes." I can live with that. What's the second phrase?
"Resulting in a material misstatement." It's true, but it's not the whole story. The misstatement never reached the accounts. We found it on day three and corrected it on day four, before we signed anything.
Graham does not say the phrase is wrong. He says it is incomplete. This is the most common and most legitimate kind of wording negotiation: adding the fact that changes how the sentence will be read.
I can't remove "material". It was above materiality.
I'm not asking you to remove it. I'm asking you to add who found it, and when.
Go on. What would you write?
Alan doesn't argue in general terms. He asks for the exact words. In any drafting negotiation, the side that brings a concrete sentence usually shapes the result.
"Resulting in a material misstatement, which was identified and corrected by management before the year-end accounts were finalised."
And you'd like me to say it was disclosed to us before our fieldwork started.
I wasn't going to ask for that.
I know. That's why I'm offering it. It's a fact, and it's relevant. I said in the walkthrough that it would be in my notes. I'd rather it was in the letter.
Nadia told Alan before he opened a single file. Here he adds that fact to the finding himself, without being asked. The finding is still significant, but anyone who reads it will see how the company behaved.
Thank you. That will matter to the team more than you'd think.
Then we accept it.
One more thing, and you can say no. Could we call it an isolated issue?
I can't. Last year had the same problem. It wasn't isolated.
Fair. Withdrawn.
Graham tries one more change, and signals in advance that a no is acceptable. Alan gives a factual reason to refuse. Graham drops it in one word. Knowing when to stop asking is part of negotiating well, and it keeps the requests that were accepted credible.
Let me read the whole thing back. "Significant deficiency. Credit notes were recorded in the wrong period due to the absence of a cut-off control for credit notes, resulting in a material misstatement, which was identified and corrected by management before the year-end accounts were finalised, and disclosed to the auditors before fieldwork began."
After changes, the full text is read once more from start to finish. Small edits can change the meaning of a sentence as a whole, and this is the moment to catch it.
That's accurate.
Agreed.
Now your response. What will you do, who owns it, and by when?
What, who and by when. An auditor will reject a response that is missing any of the three. "We will improve our processes" is not a response.
Three actions. One: since the first Monday after the close, payables record credit notes by the date on the document. The written procedure was updated last week, and you have a copy.
Two: in the first week of every quarter, every credit note over five thousand is reviewed against its document date. Aisha prepares the review and I sign it off. The first one is before the first-quarter close.
Three: we've asked IT to cost a system warning when a document date falls in a closed month. If it's affordable, it's in place by the half-year.
"If it's affordable." I'd prefer not to have a condition in a response.
Fair. Then write: "We will decide on a system control by the half-year." That's something we can promise.
Nadia's third action depends on a cost nobody knows yet. Graham rewrites it as a promise to decide, which the company fully controls. In a management response, every action should be something you can be tested on.
That I can test. And the owner of all three?
The Financial Controller. Me.
When I test the quarterly review, what will I actually see?
A list of every credit note over five thousand, with the date on the document and the date we recorded it. Aisha's initials as preparer, mine as reviewer, and the date we each signed.
And if Aisha is away in the first week of a quarter?
Then I prepare it and Graham reviews it. The rule is two different people, not two particular people.
Alan tests whether the control depends on one person being present. Nadia's answer states the principle behind it in one sentence. A control that only works when everyone is at their desk is not a control.
Put that sentence in the procedure. It's exactly what I'd want to see.
Alan asks what the evidence will look like before the control has run once. Nadia's answer is concrete: what is listed, who signs, and when. If you can describe the evidence, the control is real.
I'll test the quarterly review at the interim audit. If it's working by then, I'd expect to close this point next year.
When is the interim audit?
July. By then you'll have run the review twice, at the end of March and the end of June. I'll test both.
Then we have two chances to get it right. One will do.
Graham asks when the promise will be tested, and Alan gives a month and the evidence he will look at. Knowing the date of the next check turns a commitment into a deadline.
There's one more point, and it's smaller. Aisha told me payables classify documents by the supplier's label. If a supplier calls something a credit note, it's recorded as one.
How are you rating it?
As an observation, not a deficiency. Nothing went wrong because of it this year. But it's a gap.
Then we won't argue with it. Nadia?
For anything over five thousand, payables will check that the document matches its label before they record it. I'll add it to the same procedure, this month.
Owner and date in one sentence. Thank you.
Who else will see the letter?
The audit committee, at their next meeting. I'll present it, and you'll both be there to answer questions.
Then the response should be in Nadia's words, not mine. She'll be the one answering for it.
Graham hands the response to the person who owns the actions. It is a small sign of trust, and a practical one: the person who will be asked about a promise should be the one who wrote it.
You'll get the draft letter by Friday. You have two weeks to comment, but I don't expect much.
Neither do I. We've just done the commenting.
For what it's worth, this is the cleanest version of a bad finding I've written this year.
A modest way to introduce a personal opinion. Alan is not obliged to say it, which is exactly why it means something. It also tells Graham and Nadia that the way they handled the problem was noticed, not only the problem itself.
We'll take that. It's not a sentence I want to hear twice.
Next year, I'd like your letter to be very short.
So would I. It's the best kind to write.