Reconcile the report as a versioned bridge, not as a replacement for the underlying records. Start by defining the exact periods, entities, account groupings, adjustment definitions and scope covered by the quality-of-earnings, or QoE, work. Then map each QoE finding to the management schedule and the current presentation schedule, preserving the original amount, source, reviewer and explanation for every difference. Differences should be classified as scope, period, mapping, calculation, presentation or unresolved evidence issues. AI can help an associate locate cited items, normalize labels and draft an exception list, but a responsible financial reviewer must verify each tie-out and decide what remains open. The bridge is a diligence workpaper, not an independent QoE opinion or valuation conclusion.
Start with three records, not one blended file
The first control is to keep three records distinct. The QoE report contains the provider's findings, definitions, procedures and limitations. Management financials contain the company's books, schedules and explanations. The current presentation schedule contains the numbers being used in a management presentation, teaser or other sale-process material. A bridge connects them; it should not silently replace one with another.
Assign a version, date, period end, currency, unit and preparer to each record. Capture whether a figure is reported, adjusted, estimated, reclassified or supplied by management. If a workbook was refreshed after the QoE report, retain the earlier version and record the refresh as a new input. This prevents a later change from looking like a conclusion that the QoE provider made earlier.
Define the comparison before calculating differences
A difference has meaning only after the comparison basis is fixed. Write down the entity perimeter, fiscal calendar, cut-off date, revenue and expense definitions, account hierarchy and presentation purpose. A QoE analysis may focus on sustainable earnings, while a management schedule may include all legal entities or a different trailing period. A presentation schedule may also group accounts for readability rather than follow the chart of accounts.
Use a bridge row for each comparable measure. Useful fields include source label, source locator, period, original amount, mapped amount, adjustment category, explanation, owner and status. Keep scope and period differences separate from true numerical discrepancies. For example, a full-year management amount cannot be treated as an error merely because the QoE schedule covers a different cut-off. The reviewer should decide whether the row needs a new calculation, a disclosure, or no change.
Classify every variance instead of overwriting it
A practical classification is scope, period, mapping, calculation, presentation and evidence. Scope means the populations differ, such as entities, locations or revenue streams. Period means the dates or fiscal calendars differ. Mapping means the same underlying item sits in different account or line-item groupings. Calculation means a formula, sign, unit, currency or aggregation needs checking. Presentation means the number may be correct but displayed in another subtotal. Evidence means the supplied support is incomplete, contradictory or not yet reviewed.
This classification turns a large unexplained variance into a queue of reviewable questions. Preserve the original values beside any proposed bridge value. Do not force a difference into a recurring or non-recurring bucket solely because the final presentation needs a clean answer. SEC staff guidance on non-GAAP measures illustrates why adjustment definitions and period-to-period consistency matter: an adjustment can be misleading in context, and inconsistent treatment may require explanation or recasting [1]. The guidance is not a universal private-company QoE rule, but it is a useful discipline for documenting what changed and why.
Treat management schedules as evidence to test
Management schedules are valuable inputs, but a reviewer should test whether they are complete, accurate and sufficiently detailed for the bridge. Tie totals to the relevant trial balance or general ledger extract where available. Check formula cells, sign conventions, hidden rows, duplicated accounts, period filters and manual overrides. Trace unusual adjustments to invoices, contracts, payroll records, tax filings or other permitted support, depending on the item and the agreed diligence scope.
The PCAOB's audit evidence standard is written for audits, not QoE reports, yet its evidence concepts are useful here. It distinguishes information that supports management assertions from information that contradicts them, and it treats relevance and reliability as separate questions. For company-produced information, it calls for evaluating accuracy, completeness and precision or detail [2]. A deal-team bridge should apply that logic proportionately and label the work as reconciliation, not as an audit or independent assurance conclusion.
Make the bridge explain sustainable earnings without pretending to value the business
A QoE adjustment usually asks how a historical result should be interpreted for a defined analytical purpose. The bridge should therefore state the adjustment's source, rationale, period effect and treatment in the current schedule. Keep recurring, unusual, owner-related, timing, accounting-policy and run-rate questions distinct when the underlying evidence supports those distinctions. If evidence does not support a conclusion, mark the item open rather than converting uncertainty into a normalized number.
Do not use the bridge to infer a purchase price, valuation multiple or buyer conclusion. The same adjustment can have different implications depending on the transaction structure, diligence scope and information available. A clean reconciliation improves traceability, but it does not make a QoE report a valuation opinion. The financial reviewer remains responsible for deciding what can be stated and what requires management clarification or specialist review.
Use AI for retrieval and organization, not judgment
A bounded AI pilot can assist with low-discretion tasks. Give it a controlled set of documents and ask it to extract cited page references, identify similar labels, compare period headers, propose candidate mappings and draft a variance register. Require every proposed row to retain the source filename, page or cell locator, original text and confidence note. The output should be suggestions for an associate, not a new source of financial facts.
Set explicit stopping conditions. Stop the pilot if citations cannot be reproduced, if the tool invents a source locator, if it merges different periods, if it drops negative signs or units, if it changes an adjustment definition, or if the reviewer cannot explain the result from the source records. A human financial reviewer should approve each accepted mapping and adjustment. AI should not contact management, send a request, approve a disclosure, negotiate an issue or decide that a QoE conclusion is correct.
Protect confidential information and document the pilot
Before uploading deal materials, identify the confidentiality class, permitted users, retention terms, model-training settings, storage location, deletion process and incident contact for the selected tool. Use the least sensitive extract that can answer the pilot question. Mask personal information when it is not needed. Keep the original records in the approved workspace and store the AI output as a draft with access controls, not as the system of record.
NIST describes its AI Risk Management Framework as voluntary guidance for managing AI risks and incorporating trustworthiness into the design, use and evaluation of AI systems [3]. FINRA's discussion of generative AI also emphasizes asking vendors how data is used, where it is stored, who can access it, whether it is restricted to the firm's use, how failures are handled and how confidentiality is protected [4]. Those sources do not create a blanket rule for every boutique bank. They do support a documented vendor and data review before a pilot, plus a manual fallback if the tool is unavailable or its output cannot be verified.
Make the deliverable a reviewer-ready exception log
The finished bridge should let another reviewer reconstruct the path from source to conclusion. Include a reconciliation summary, a row-level bridge, a source index, scope and period notes, adjustment definitions, open exceptions and sign-off fields. For each open item, state the question, owner, evidence needed, due date if the team uses one, and what cannot yet be said. Keep a change log when a source or mapping changes.
A useful final review asks five questions. Did every accepted row tie to a source? Did the periods and entity scope match? Were original amounts preserved? Are unresolved items visible in the presentation schedule or review notes? Did a human reviewer approve the treatment? The intended implementation outcome is a human-reviewed QoE-to-management bridge with source references, scope differences and open exceptions. That outcome is stronger than a polished spreadsheet that hides the path taken to reach its numbers.
Where Acquivo fits in the workflow
A boutique investment bank can use this framework as a repeatable workpaper pattern: define the source set, create the versioned bridge, run a constrained extraction pilot, review exceptions, and only then update presentation schedules. The software should remain user-operated, with the deal team controlling permissions, source selection, review and final use of the output. Acquivo is a software platform that delivers the entire sell-side M&A process, valuation, marketing materials, buyer matching, and deal management, on one platform for business owners, M&A attorneys, and boutique investment banks.
That positioning does not turn a reconciliation into an independent QoE opinion, legal advice or a promise of accuracy. The responsible team still decides which records are authoritative for its engagement, what evidence is sufficient for its purpose, and when to consult accounting, tax or legal professionals.
Key Points
- Keep the QoE report, management financials and current presentation schedule as separate versioned records.
- Define entity scope, periods, units, account mappings and adjustment definitions before calculating a variance.
- Classify differences as scope, period, mapping, calculation, presentation or evidence issues, and preserve original values.
- Use AI to extract citations, compare labels and organize exceptions, with a human financial reviewer approving every accepted row.
- Set stopping criteria for uncited, period-mixed, sign-changing, confidential or otherwise unverifiable output.
- Deliver a bridge with source references, open exceptions and a change log, not a valuation conclusion or independent QoE opinion.
Sources and references
Source retrieval dates record research access, not article publication. Read our editorial and date policy.
- [1] Non-GAAP Financial Measures · Accessed
- [2] AS 1105: Audit Evidence · Accessed
- [3] AI Risk Management Framework · Accessed
- [4] Generative AI and Large Language Models · Accessed
Choose your next step
Acquivo is user-operated software. Owners and their professionals remain responsible for decisions, communications, and professional review.
Educational information, not individual legal, tax, or investment advice. Consult qualified professionals for your circumstances. Acquivo provides software, not representation or negotiation on your behalf.