Use ChatGPT Work to Build a Stakeholder Question Log

A project brief can look complete because it has a goal and a deadline. The missing decisions usually appear later, during the meeting when everyone is already busy.

This guide shows a documentation-based ChatGPT Work workflow for making those unknowns visible before a stakeholder review. The project brief and output are fictional. No ChatGPT Work execution is claimed.

What you will create

You will create a question log with five fields: question, why it matters, source section, proposed owner, and decision status. The log is a preparation aid, not a decision register.

Level: Intermediate. You need access to ChatGPT Work on the web or desktop surface described in the current OpenAI documentation, plus an approved project brief. Work availability and controls can roll out gradually, so confirm that your account shows the surface before using these steps.

The fictional project brief

Paste this into a text file called portal-refresh-brief.txt:

Project: Supplier portal refresh

Goal: Give approved suppliers a clearer way to check order status.

Target: Prepare a pilot for review in November.

In scope: status page, email notification copy, and a small pilot with selected suppliers.

Out of scope: changing the order database and redesigning the internal finance screen.

Known dependency: the operations team must confirm which status labels suppliers can see.

Open note: the brief does not name the pilot suppliers, the November review date, the person who approves email copy, or the success measure.

Build the question log

Open ChatGPT Work and start a bounded task using the approved brief as the only source.
Attach portal-refresh-brief.txt if file upload is available, or paste the text into the task.
Ask for a plan before the log. The plan should name the source file and the sections it will inspect.
Paste the instruction below.
Review the result against the brief. Do not accept an owner, deadline, supplier list, or success measure that the source does not provide.

Read only the project brief supplied for this task. Create a stakeholder question log before a review meeting.

Do not invent project facts, owners, dates, suppliers, approvals, metrics, or commitments. A job title is not an owner. Keep the distinction between a fact, a dependency, and an unknown.

Return:
1. Confirmed scope, in four bullets.
2. A Markdown table with these columns: question, why it matters, source section, proposed owner, decision status.
3. A short list called "Questions already answered by the brief".
4. A short list called "Missing decisions that block a clean review".

Use `unassigned` when the brief names no owner. Use `not set` when it names no date or measure. Cite the exact phrase or section that caused each question. Do not turn a proposed pilot into an approved launch.

Illustrative result

A useful log should ask who selects the pilot suppliers, who approves the email copy, which status labels are allowed, what the November review date is, and how the pilot will be judged. It should mark the operations team’s status-label confirmation as a dependency, not as a completed decision. This is an editor-written expected result, not captured model output.

Acceptance checks

The task is ready for a stakeholder meeting when:

the log keeps the status page, email copy, and pilot in scope
the database change and finance screen remain out of scope
the operations dependency is present and tied to its source phrase
missing pilot suppliers, review date, email approver, and success measure stay unresolved
proposed owners are marked unassigned where the brief provides none
every question points to a phrase or section in the brief

Troubleshooting

The log contains invented owners. Repeat the instruction with Never infer an owner from a department or role. Use unassigned.

The model turns questions into decisions. Ask for a second table with source wording and interpretation, then remove any interpretation that is not supported.

The task uses extra context. Remove unrelated files and rerun it with the single approved brief. A question log is only reviewable when its source boundary is clear.

Limits and safer use

This workflow exposes missing decisions. It does not decide them, check whether the plan is feasible, or replace stakeholder approval. Keep confidential project details in an approved workspace and verify the current access, file-handling, and retention rules for your account.

For a checkpoint-based workflow, see Use ChatGPT Work to Keep a Multi-Step Task From Drifting.

Want one useful workflow each week? Join The Weekly Verdict.

Sources

Evidence label: Documentation-based guide with fictional input and an illustrative output. No ChatGPT Work execution or personal test is claimed.