Key takeaways
- Define completeness by the decision, the screen, the questions, the attributed statements, and the retention file.
- Write the note the next morning, while testimony and impression are still distinguishable.
- Treat a refused question as a record entry rather than a gap in the conversation.
- Measure the delay from the call to the note and from the note to the decision.
Question and scope
This audit asks whether an executive office can show why an expert network call happened, what the expert was asked, what the expert said, and how the note was used. The unit is one engagement tied to one decision and one expert. The review follows the screen, the engagement letter, the question list, the note, and the invoice. It does not judge whether the expert was right.
Set the scope to one requester and a fixed period, such as two consecutive quarters. Record the decisions the calls were meant to inform and the number of engagements. If the requester changed during the period, separate the calls. A new requester can change the record without changing the process.
Include calls that produced a decision, calls that produced nothing, and at least one call where the expert contradicted the office's prior view. Record the category and the outcome. A sample of only useful calls will overstate the quality of the record. The sample should also name the decision each call was meant to inform.
Include a call where the expert was screened out after the first conversation and one where the network changed the expert at short notice. A substitution is a record event, not an administrative detail. Record the timing and the reason. The audit should test whether the file reflects the person who actually spoke.
What the sources establish
The SEC's public material describes the general prohibition on trading on material nonpublic information, which supports treating expert calls as a controlled activity without defining an executive office record. GAO's Green Book describes control activities and documentation as parts of an internal control system, which supports a record-based review. BLS occupational data describes the assistant's coordination role but not an expert call standard. The sources do not establish what an expert call note must contain.
The sources do not set a retention period for call records or a required level of attribution. They also do not say whether an expert must be screened against the company's counterparties. The audit therefore measures the office's own engagement policy and the file it produces. The report should keep that local scope visible.
The sources support separating the record of testimony from the record of decision. A note can capture a strong statement and still leave the decision unclear. The audit should test both elements. A complete-looking note is not evidence that the boundary was respected.
Definitions and measures
Define a complete expert call record as one with the decision, the requester, the expert's identity and role, the conflict screen, the questions, the attributed statements, the declined topics, the consent for any recording, and the invoice. Keep the field list fixed across the sample. A note with an unattributed summary is counted as incomplete even when the call was useful. The same list should apply to every engagement in the period.
Measure the days from the call to the note and from the note to the decision. A note written two weeks later is less useful than one written the next morning. Report both intervals. The measure shows whether the record supported the decision it was meant to inform.
Record whether each substantive statement is labelled as observation, opinion, or hearsay, and whether the note names the source. A conclusion written in the company's voice is a different artifact from testimony. Report the three labels separately. Keep that field beside the question it answers.
Record whether the expert's conflict screen was current for the call date. A screen from the previous year may not cover a new employer or a new deal. Report the age of the screen. The audit should treat currency as part of the record.
Sampling and collection
Sample calls from the network invoices, the requester's calendar, and the office's research folder. Include a call that was cancelled and one that was rescheduled. Record the source and the category. An invoice-only sample will miss the calls that were never billed or never happened.
Collect the engagement letter, the screen, the question list, the consent, the note, the recording if any, and the invoice. Mask personal details that are not needed. If a recording exists, record its storage location and access list. A recording without a note is still an incomplete record.
Include a call where the expert asked for the questions in advance and one where the questions were sent after the call began. The difference affects how much of the answer is spontaneous. Record the timing and the list. The audit should not treat every call as prepared in the same way.
Include a call made by a colleague outside the executive office and later filed with the office. A decentralized call is the most likely to lack a screen and a note. Record the origin and the file. The audit should test the boundary of the office's control.
Verification pathway
Check the note against the question list and confirm that every question has an answer, a refusal, or a note that it was not reached. A gap that is not explained looks like an omission. Record the question and the status. The audit should test coverage, not only length.
Confirm that the attributed statements match the recording or the contemporaneous notes where one exists. A paraphrase that changes the direction of a statement is a serious finding. Record the comparison and the result. The audit should include at least one call with a recording.
Check the invoice against the engagement letter, the booked duration, and the approved rate. A call that was extended without approval is a control finding even when the note is good. Record the variance and the approval. The audit should test the payment, not only the prose.
Confirm that the file can be retrieved by the decision reference rather than only by the expert's name. A file that is searchable only by a person is hard to use after a role change. Record the reference and the retrieval test. The audit should include a file opened by someone who did not make the call.
Analysis and reporting
Report the number of calls, the completeness rate, the median note delay, and the share of calls with a current screen. Break the results down by requester and by decision type. Report the calls with no decision separately. A single completeness rate can hide a set of calls that consumed budget without informing anything.
Inspect the gaps for patterns. A cluster of missing screens may point to the network's onboarding, while a cluster of late notes may point to the reviewer. Choose one change, such as a screen checklist or a next-morning note deadline. Do not rank assistants by note quality when their call mix differs.
Report the days from the note to the decision, not only the call date. A note can be timely and the decision still delayed for unrelated reasons. Show the median and the longest case. The measure reflects the use of the record rather than its existence.
Limits and next action
The audit cannot prove that a complete file prevented a bad decision or that every statement was heard correctly. It also cannot capture a conflict that emerged after the screen. The report should state those limits and avoid treating the note as a transcript. Keep the screen and the questions with the study so the next reviewer can check the same record.
Set a repeat audit date and keep the field list stable. Assign the correction to the research owner and give the requester a one-page call checklist. The next quarter should test whether the same screen or the same note delay fails again. A useful result is a call that is documented before the decision is made.
Keep the audit record with the engagement files so a later review can compare the claim with the evidence. Set the next sample date and name the owner. A research routine that is not reviewed will lapse with the next role change. The office should be able to show what improved.
Sources
- Secretaries and Administrative Assistants, U.S. Bureau of Labor Statistics, Occupational Outlook Handbook.
- Executive Secretaries and Executive Administrative Assistants, U.S. Bureau of Labor Statistics, Occupational Employment and Wage Statistics.
- Executive Secretaries and Executive Administrative Assistants, O*NET OnLine.
- Business Formation Statistics, U.S. Census Bureau.
- Nonfarm Business Sector: Labor Productivity, Federal Reserve Economic Data.
- Productivity Statistics, OECD Data Explorer.
- World Development Indicators, World Bank DataBank.
- ILOSTAT Labour Statistics, International Labour Organization.
- 2024 Work Trend Index Annual Report, Microsoft and LinkedIn.
- The economic potential of generative AI, McKinsey Global Institute.
- Creating helpful, reliable, people first content, Google Search Central.
- Search Engine Optimization Starter Guide, Google Search Central.
- Dear Manager, You Are Holding Too Many Meetings, Harvard Business Review.
- U.S. Securities and Exchange Commission.
- GAO, Standards for Internal Control in the Federal Government (Green Book).
- BLS, Secretaries and Administrative Assistants.
Related research
Apply the research
Use the finding to frame one question for your own team: which support work is currently invisible, and who should own the next step? A scoped consultation can turn that question into a practical role. Write the question down before the meeting. Name the person who owns the next step.
Turn the finding into a support scope.
Bring the workflow that feels hardest to see. We will help define the role, the evidence it should maintain, and the decisions that stay with leadership.
Talk through the role