Audit Preparation: Arriving at Fieldwork With Numbers You Don't Have to Defend

The external auditor doesn’t approve your purchase of anything and doesn’t sign off on your systems. But no one else in the room benefits more directly from whether your close is actually defensible, and no one is better positioned to make fieldwork longer if they don’t trust what they’re handed. An auditor who distrusts a client’s numbers adds procedures, asks for more, and extends the engagement, quietly or not so quietly. An auditor who has seen a clean trail becomes the easiest reference a CFO has for “does this actually hold up.”

What the auditor is actually checking

Most finance tools that claim “audit-ready” mean drillback to the general ledger line. That’s useful, and it’s also table stakes: most FP&A platforms already do it. What an auditor actually wants, and what most tools stop short of, is drill-to-transaction: a number that resolves not just to a GL entry but to the subledger behind it and, ideally, the actual source document (the contract, the invoice, the bank statement) that proves the entry is real. That last hop is the one most tools don’t make. It’s the difference between “the client says this reconciles” and “here is the document that proves it reconciles,” and an auditor who has reviewed enough PBC packs and roll-forwards can tell the two apart immediately.

Where this actually matters on the calendar

Audit pressure isn’t evenly distributed across the year. It concentrates around the annual filing cycle and around any change in audit status: crossing a small-company exemption threshold, or switching auditors. In the run-up to that filing, the months before fieldwork are when a close being defensible actually gets tested, not the weeks after. That’s the window where a provenance trail that already exists is worth more than one you’d have to reconstruct from someone’s memory of what happened three quarters ago.

The gate that keeps a bad number from reaching that stage at all

The mechanism behind this is a verification gate that sits between a model and anything it produces. Before a board pack, workbook, or statement can be exported, the model’s own statement identities get rechecked: does the balance sheet actually balance, does the cash-flow statement’s net change match the balance-sheet movement it’s supposed to explain. A break on an actual, source-anchored period is a hard block: no export, and the specific broken identity is named rather than papered over. A break on a forecast period is advisory only, because a forecast is allowed to be provisional in a way an audited actual is not. The checks are deterministic (same model state, same verdict, every time), and every export path in the product runs through the identical decision, so there’s no second, looser version of the rule quietly living somewhere else. In practice, that means a document leaving Rexfin either ties out or it doesn’t leave.

Citability is checked the same way. A figure only counts as verified if it’s backed by independent extraction agreement on its source value: a number nobody can trace to a page in a document doesn’t get to count as reconciled just because the arithmetic happens to work.

What this changes for the monthly close

None of this replaces the judgment an auditor applies, and it shouldn’t try to. What it changes is the starting point of the conversation. Instead of walking into fieldwork with a workbook and a promise that everything reconciles, a controller can hand over a pack where every material figure already carries its source, and where anything that didn’t verify was blocked from export rather than shipped with a caveat nobody reads. The controls behind this (no training on customer data, access designed to enterprise-grade standards) are the same posture an auditor’s own risk team would ask about before trusting any tool with the numbers in the first place.

Who this is for

This is built for the finance teams who feel the audit six months before it starts, not the two weeks before fieldwork: controllers preparing a year-end close, CFOs fielding audit fieldwork requests they can’t fully predict in advance, and anyone who has had to explain, after the fact, why a number in last year’s pack didn’t match what the ledger actually said. It’s also directly relevant if you’re assembling a due diligence data room, where the same “show your work” standard applies under a tighter clock. See the rest of what a defensible close looks like at the finance team use cases hub.