A written answer on whether your subsidiary's numbers can be trusted

From head office a subsidiary is a line in the consolidation and a person who answers emails slowly. When the numbers stop making sense, the group has no way to find out why. The local team reports on the very thing being questioned, the auditor speaks once a year and only about the past, and flying a group controller in for two days produces a polite tour and no answers.

What is missing is somebody senior who can sit in the finance room, read the ledger in the local language, ask the second and third question, and then write down what they found for the people paying for it. That is all this is.

The order matters. Each answer decides which question is worth asking next, and a review that starts at the interesting end usually never reaches the ledger.

• Does the ledger tie to the last signed accounts

  The opening position of everything else. If the local books do not agree with the last audited financial statements, nothing built on top of them means anything, and every later finding is provisional until this is closed.

• Are the bank accounts reconciled, to what date, and by whom

  Not whether a reconciliation exists, but whether it was performed by someone other than the person making the payments, and what sits in the unexplained line. This is the single fastest read on whether a finance function is in control.

• Are the receivable and payable ledgers real

  Ageing that nobody has challenged in a year is decoration. We test whether the oldest balances are collectable, whether credit notes are being used to tidy the ledger, and whether anyone has told the group what is genuinely at risk.

• Does every intercompany balance agree with its counterparty

  Entity to entity, both directions, at a date. Differences carried as a plug for years are common, and they are the item most likely to surface in a due diligence or a tax examination rather than in a close.

• What has actually been filed, and what is open

  Tax returns, statutory accounts, registry filings, licence conditions. What was filed, when, what is late, and above all whether anything is open with a clock running on it. Assessments in this region come with response windows measured in days.

• Where the entity holds incentives, whether they would survive an examiner

  Promoted and non-promoted activity separated in the books, a cost allocation method that is documented and consistently applied, and the annual incentive reporting actually filed. Designed at the start this is routine. Reconstructed three years later it is expensive, and the sanction at the end of the road is loss of the incentive.

• How long the close really takes, and how much of it is estimated

  The reported close date and the real one are often different. We look at what is accrued on judgement, what is reversed the following month, and how much of the pack is assembled outside the accounting system.

• What it runs on, and whether the statutory outputs come out of it

  Whether the local entity's tax and statutory forms are produced by the system or rebuilt by hand, whether the system is registered where local rules require it, and what the group's own reporting deadlines are doing to the local team.

• Who does what, and what happens if they resign

  The org chart as it is rather than as drawn, the single points of failure, who holds the bank tokens and the tax portal credentials, and the notice periods. In most subsidiaries of this size the answer is one person, and the group does not know their name.

• The gap between the local books and what group receives

  Every adjustment made between the statutory ledger and the group submission, who makes it, and whether it is documented. This is where local reality and group reporting quietly separate, and it is the hardest gap to close retrospectively.

Read access to the accounting system, the last two signed audit files, the filing history, twelve months of bank statements, the group submission pack and the finance team org chart.

Asking for exactly that list is itself part of the review. A group that can produce it within a week has a subsidiary with a paperwork problem. A group that cannot has something else, and we will have learned it before the engagement has properly begun.

We also ask the parent to tell the local team directly that the review is happening and who commissioned it. Reviews introduced as a friendly visit go badly. Reviews introduced honestly are usually welcomed by the local finance manager, who has often been asking for help for a year.

• Findings ranked by money

  What each issue is worth, sized rather than described. An unreconciled bank account is not a finding until somebody says how much of it is unexplained.

• The same findings ranked by deadline

  A different order, and the one that forces action. Some things cost more but can wait. Others are small and expire in a fortnight, and those are the ones that turn into penalties.

• A remediation sequence

  What to fix first, what depends on what, an estimate of time and cost against each step, and who can realistically do it: the existing team, the group, us, or a local specialist.

• A one page summary

  Written so the group CFO can forward it to their own board or their auditor without editing it first. If a report needs translating before it can be used, it was not finished.

It is not an audit, it is not a tax opinion, and it is not a forensic investigation.

We are not auditors and this review gives no assurance. Where a finding turns on a local statutory position, it goes to a local tax adviser or to counsel, named in the report, and we say plainly which of our conclusions depend on their confirmation. In our six core jurisdictions we can usually take that position ourselves. Outside them we hold the group view and the reporting discipline and we rely on a local specialist for the statutory detail, and we would rather write that down than pretend otherwise.

If what we find looks like fraud rather than disorder, we stop and tell you. That is a different engagement with different rules, and running it as an extension of this one would damage it. Forensic work is scoped separately.