Data Analytics in Auditing, What It Changes

Data analytics tests all 40,000. That is the whole change, and everything else follows from it.

Why sampling existed

Not because it was better. Because examining every transaction by hand was impossible.

Sampling is a workaround with real weaknesses. It finds patterns only if they are common enough to appear in your sample. A scheme running once a month across a year will usually be missed entirely.

What full population testing finds

Six tests that are trivial on complete data and near impossible on a sample.

1. Every journal posted outside working hours, or on a weekend.
2. Every payment just below an approval threshold. A cluster at AED 4,800 against a AED 5,000 limit is not chance.
3. Duplicate payments. Same amount, same supplier, close dates.
4. Round-number transactions, which correlate with estimates and manipulation.
5. Supplier bank details matching employee bank details.
6. Journals posted by people whose role should not allow it.

None of these require artificial intelligence. They are queries over your ledger, and they find real things.

What it means for you as the client

Better questions, and more specific ones.

An auditor working from a sample asks broad questions. One working from the full population arrives with a list. Twelve payments to this supplier, all just under your limit, explain these.

That is uncomfortable and it is worth more.

It also means the data you provide matters more than it used to. A complete, clean general ledger extract enables all of this. A partial one disables it.

The limits, stated honestly

Analytics finds anomalies. An anomaly is not a finding. Most exceptions have ordinary explanations, and someone still has to look at each one.

It cannot see what was never recorded. Cash taken before it entered the system leaves no trace in the ledger.

It cannot judge whether an estimate is reasonable. That remains judgment.

And it does not reduce the audit to a script. It moves auditor time from selecting samples to investigating exceptions, which is where the value was always going to be.

What to ask your auditor

Two questions.

Do you test the full population or a sample? If it is a sample, why, given your ledger is already a data file?

What analytics do you run as standard? A firm that runs the six tests above will say so specifically. A firm that says it is "technology enabled" is describing its marketing.

Where we fit

Our audit fieldwork is software driven. We work from your complete ledger extract rather than a hand-picked sample, which is also why our document requests are specific rather than open-ended.

Send your trial balance and we will tell you what the data shows before we quote.