Evidence that answers for itself
What separates a document from support: structure, provenance, and context.
Why the teams with the calmest audits treat readiness as standing infrastructure — and what that looks like in practice.
Most finance teams experience the audit as a season. Fieldwork approaches, the request list arrives, and for six weeks the close calendar bends around evidence-gathering. Then it ends, everyone exhales, and the whole apparatus is dismantled until next year.
The teams with the calmest audits do something different. They treat readiness as standing infrastructure — a system that runs all year, quietly, as part of how the work is already done. This article describes that system: what it replaces, what it is made of, and how to start building it without adding headcount.
Seasonal readiness fails for a simple reason: the evidence an audit needs is produced all year, but it is only assembled at the end. By the time the request list arrives, the people who did the work have moved on, the context has faded, and reconstruction begins.
The symptoms are familiar:
None of this reflects a control weakness. It reflects a storage and retrieval problem — the work was done, but it cannot answer for itself.
Standing readiness has one design principle: evidence is captured where the work happens, at the time it happens. Everything else follows from that.
A reconciliation is evidence the moment it is prepared — if it is stored with its inputs, its preparer, and its date. The same document rebuilt eight months later is a reconstruction, and auditors can tell the difference. Standing readiness means the file that supports an answer is the file that produced it.
Every recurring request has a named owner and a known source before the audit begins. When the query arrives, it routes to the owner; it does not queue behind a single inbox.
The calmest audits belong to teams that never stopped being ready.
In practice, standing readiness is a loop, not a project. It has four movements, repeated every cycle:
Every audit opens with a predictable set of requests. If each one has a standing source, the first week of fieldwork becomes an export, not an excavation.
| Request | Typical owner | Standing source |
|---|---|---|
| Trial balance and mapping | Financial controller | Close package, exported at period end |
| Bank reconciliations | Accounts team | Filed with statements at each close |
| Revenue recognition memo | CFO / controller | Policy library, versioned |
| Access and change logs | Systems owner | Quarterly export, retained |
| Board minutes | Company secretary | Governance folder, indexed by date |
The table is deliberately boring. That is the point: readiness is mostly the discipline of deciding, once, where each answer lives — then keeping that decision true.
A structured record changes what software can do with an audit query. Audit Intelligence™, the reasoning layer inside Oditable, reads each query, locates the relevant evidence in your record, and drafts a response with its support attached — citations first, prose second.
In practice
An auditor asks why deferred revenue moved 40% year on year. Audit Intelligence™ drafts the response from the recognition memo, the contract schedule, and the close notes — and links each claim to its source. A controller reviews the draft, adjusts one sentence, and approves. Elapsed time: minutes, not days.
Nothing leaves without review. The system prepares; a person approves. That order never reverses, because an audit response is a representation of the business — it should carry a human signature, with the machine doing the assembly.
Standing readiness is built incrementally. A reasonable first quarter looks like this:
The first cycle is work. The second is noticeably lighter. By the third, the audit has stopped being a season — it is simply a period in which the system gets read.
What separates a document from support: structure, provenance, and context.
Controls exist to make assurance cheap. Most teams design them backwards.
The reasoning layer behind every draft — and why it cites before it writes.