A recommendation that holds up to another view.

Start with the choice the client faces

Name the decision, the two credible alternatives, and the constraint that matters. A specific brief helps distinguish an objection that changes the recommendation from one that merely adds caveats.

Use material you are authorized to share with the models. You can try the workflow with an anonymized or fictional brief before bringing real client work into it.

A question to start with

Follow the disagreement that matters

If the responses favor different options, identify the assumption behind each. Is the difference about cost, execution capacity, time horizon, or the evidence itself? Ask the follow-up that connects that assumption to the client’s situation.

When several responses raise the same objection, give it a place in your review. Then decide whether the evidence supports it. Consensus is a signal to examine, not a substitute for your judgment.

Put the reasoning in the deliverable

A client needs the recommendation and the reason for it. Select useful passages from a response and place them into the Quartz document. Edit them alongside the surrounding evidence and qualify uncertainty where it affects the decision.

Review the exported Word document or PDF before sending it. A polished deliverable should preserve the argument, its limits, and the next action.

Test the process on one memo

Use a past recommendation whose outcome you understand. See whether another perspective identifies a risk you would have wanted to discuss. That is a concrete way to evaluate the workspace before changing your process.