QC 1000 belongs in the business case

Approving an artificial intelligence tool and deciding that an audit team can work with fewer people are different decisions. Evidence supporting the first does not automatically support the second. Before an audit firm changes the resources supporting its engagements, the people approving the change should explain which quality assumptions must hold and how they will find out whether those assumptions remain sound.

Processing Content

With the Public Company Accounting Oversight Board’s QC 1000, A Firm’s System of Quality Control, scheduled to take effect on Dec. 15, 2026, firms performing engagements under PCAOB standards have an opportunity to connect implementation with business approvals. I propose a short quality-impact statement within the existing business case for consequential changes. It would record the quality objectives affected, the resources and responses needed, the evidence supporting the decision, and the conditions for reconsidering it. This is a proposed management practice, not an additional PCAOB form or a substitute for complying with the standard.

The foundation already exists. QC 1000 connects strategy and resource allocation with quality and requires timely responses to relevant changes (QC 1000, paragraphs .22-.25). The proposal builds on that continuing process. The Sept. 9 amendments were just approved today by the SEC. This proposal concerns the continuing relationship between business decisions and quality risks, rather than provisions governing particular reporting dates or organizational arrangements.

The accounting perspective is straightforward: a resource commitment, a forecast benefit and an observed result are different things. Keeping them separate would help leadership see whether a commercial proposal depends on quality safeguards that are funded and available, still being developed or merely assumed. The purpose is to make the conditions of approval usable by the people who must implement and monitor the decision.

The starting question would be what changes for an engagement or for the firm’s QC system. A new technology, shared-service arrangement, practice acquisition or change in specialist availability could warrant attention. These are examples, not a presumption that such decisions reduce quality. A better tool or access to deeper expertise may improve performance. The relevant trigger is a meaningful change in quality risk or in the resources and responses supporting the work, rather than the size of the invoice alone. A low-cost change can alter a consequential procedure.

For a decision meeting that threshold, the statement could begin with the intended benefit and the proposed operating change. It would link each important assumption to an existing quality objective and risk assessment, identify the people or engagements affected, and explain how the firm expects to maintain or improve quality. A reference to existing documentation would be preferable to copying a risk register into another file.

The resource plan needs the same specificity. If the change depends on specialist consultation, reviewer capacity, training or access to a provider’s records, someone should confirm that support is available when needed. QC 1000’s resources objectives address competence, time, technology and external resources (paragraphs .43-.44). The proposed statement would connect those considerations to the particular approval. It should identify who can authorize additional support or restrict the change if an essential assumption fails.

A hypothetical firm illustrates the distinction. It plans to use AI to extract terms from contracts for audit teams and forecasts fewer preparation hours. Approval to test that tool would not also authorize reducing the personnel available for exception analysis. The pilot would first need to show what happens when documents are incomplete, contract amendments conflict, or the system produces an uncertain answer. These are proposed test conditions, not reported results from a particular product or firm.

The business case should assess whether the apparent saving survives the work of checking sources, resolving exceptions and documenting conclusions. A reduction in preparation time accompanied by an increase in specialist review may still be worthwhile. But the staffing decision should reflect where the work moved. A pilot can support further use of a tool without yet supporting the withdrawal of existing review capacity.

For that hypothetical project, the statement could identify the permitted use, the engagement types included, the source records retained, and the process for handling unsupported outputs. The PCAOB staff’s implementation guidance highlights integrity and reliability for AI tools used in engagement decisions (QC 1000 Questions and Answers, VII.B.3). The firm could assess those characteristics before expanding deployment, then consider a separate resource change using the results. The proposal is to separate two approvals that depend on different evidence, not require every project to retain its original staffing indefinitely.

The same reasoning applies without AI. In a second hypothetical, a firm consolidates technical consultation into a shared team. Deeper specialization may be a benefit, but the decision should explain coverage during overlapping deadlines, escalation when a specialist is unavailable, and access to a workable alternative. If the fallback uses the same people already committed to the central team, its availability requires examination. The record should distinguish a smaller local budget from a reduction in the total work required across the firm.

That distinction also matters when several changes occur together. A training reduction, new engagement commitments and a revised review model might each appear manageable on its own while depending on the same experienced personnel. The approving group should consider the combined demands before crediting the same capacity to several proposals. The quality-impact statement is most useful where it exposes a shared assumption that separate budgets leave unclear.

Responsibility for that explanation should remain with the people sponsoring and approving the business change. The quality function can challenge the assumptions and advise on responses without becoming the sole owner of a commercial decision. QC 1000 places ultimate responsibility for the QC system with the firm’s principal executive officer (paragraph .11). A project-specific record should support, not redistribute, that responsibility. Monitoring must also respect the standard’s objectivity requirements; generally, individuals cannot monitor their own work (paragraph .61, note).

Approval should identify what must be demonstrated before implementation, what remains uncertain during a bounded pilot, and what would require restriction or redesign. A review date alone is insufficient when an essential condition can fail sooner. Loss of an agreed specialist, a consequential tool change or expansion to unfamiliar engagement types could trigger reassessment. A conditional approval would not authorize noncompliance with professional requirements; it would define the conditions within which the proposed change may proceed.

Once the decision is implemented, the original assumptions should remain visible beside observed results. Useful questions include whether promised support arrived, whether exceptions were resolved as planned, and whether workload shifted to people omitted from the original proposal. Financial and quality results should be evaluated separately. A project could miss its savings target while improving quality, or reduce recorded hours while leaving important work unresolved. Neither conclusion follows from an hours report alone.

A disappointing result should prompt investigation rather than an automatic verdict against the technology or the engagement team. Was the operating design unsuitable, the information incomplete, the support unavailable, or the original estimate unreasonable? The record can help the firm locate the decision that introduced the condition, while avoiding hindsight assumptions that every adverse outcome was predictable. It should also preserve evidence of genuine improvements and the circumstances that made them possible. Indications of quality problems would enter the firm’s existing evaluation and remediation process (QC 1000, paragraphs .71-.76), rather than being treated only as project overruns.

There is a reasonable objection: firms already have risk assessments, approval processes, and monitoring records. Another document could consume time without changing a decision. The answer is integration, not a parallel compliance system. A section in an approval memorandum, with links to the relevant QC documentation, may be sufficient. PCAOB staff guidance explains that similar technologies may be grouped and shared policies used where appropriate (Questions and Answers, VII.B.1-VII.B.2). This proposal should be equally proportionate and should not require a separate statement for every routine purchase or assignment.

Firms could begin with a forthcoming consequential decision, asking those responsible for quality, technology, staffing and engagement delivery which assumptions they would need to verify. They could then assess whether the record changed the approval, secured missing resources, clarified a restriction or improved subsequent review. Those are proposed ways to evaluate the method, not evidence it has already improved audit quality. A firm whose existing process makes these connections clearly has no reason to duplicate it under a new name.

The resulting evidence could inform the annual QC evaluation, but the statement would neither replace that evaluation nor demonstrate the effectiveness of the entire system. Its value would be more immediate: making the conditions behind a business decision visible while leadership can still change them. Audit quality should be part of deciding what resources to commit, what work to change, and what support can safely be withdrawn. A firm should be able to explain not only why it approved the change, but what evidence continues to justify it.