A framework for evidence completeness

No data room is complete in the abstract. It is complete or incomplete relative to a question. A ledger covering three fiscal years is complete for a trend in gross margin and incomplete for a trailing-twelve-month EBITDA that ends last month.

State the question first, then list the records it needs, then compare that list with what has been received. The difference is the evidence gap, and it should be reported in the same place as the finding.

Treat every finding as carrying a coverage statement: which entities, which accounts, which months, and which document types it rests on. A reader who can see the coverage can judge the finding.

• Entity coverage. Every legal entity in the transaction perimeter, including dormant ones and any the owner uses for rent, payroll, or purchasing.
• Period coverage. Every month of the analysis period, with no gaps at year ends or at the most recent month, where the run rate is measured.
• Account coverage. Every bank, credit card, loan, and merchant account, including closed ones, and every ledger account mapped to a statement line.
• Document coverage. The contracts, filings, and schedules that explain the material balances: top customer contracts, leases, debt agreements, tax returns, and the addback support.

Score each dimension for each finding rather than for the engagement as a whole. Cash proof may be fully covered for the operating account and uncovered for payroll; revenue analysis may be covered for the parent and uncovered for a subsidiary.

Preserve the identity of the source file, the relevant row or document location, the entity, and the analysis period. Similar filenames or repeated exports are not enough to establish that two reviewers are looking at the same evidence.

Normalization should make records easier to analyze without erasing their origin. Mapping a ledger account into a financial category is a transformation that should remain understandable to the person reviewing the result.

Distinguish source values from calculated values. A financial figure copied from an accounting export is not the same kind of evidence as a metric derived from several transactions or a proposed management adjustment.

For a material calculation, document the inputs, period, treatment of missing values, and relevant exclusions. A total that reconciles does not by itself establish that the selected inputs are appropriate.

Evidence gaps should remain visible through handoffs. If bank statements cover only part of the analysis period, a downstream summary should not imply that a full-period cash proof has been completed.

Exports should communicate the material assumptions and limits of the work. A reader should be able to distinguish a screened question, a reviewed adjustment, and an unresolved item.

Candidate addbacks need both numerical support and a rationale. Retain the original expense, proposed adjustment, supporting documents, and reviewer decision rather than overwriting the historical record with an adjusted result.

In Finsider Platform, the data room's outstanding-items list is the coverage statement for the engagement, and a check that lacks support is reported as not assessable. Addback retains source hashes and row citations in its derived reports. This note describes an evidence framework for discussion and implementation. It does not claim a measured error reduction, certification, or completed external validation.