Can someone else retrace the reasoning behind your conclusion?
A strong brief is shorter than the evidence behind it but does not conceal the evidence's limits. It connects a decision to inspected records, distinguishes facts from interpretation and states what additional work would change the answer. Build it from actual tool results, not a model-written replacement dataset or a list reconstructed from memory.
Write this brief from the actual inspected result: decision and scope; supported findings with source links; counterexamples; actual counts and source gaps; interpretation versus observation; unresolved questions; one bounded next step. Do not invent missing evidence or perform the recommended actions.Restate the original question
Open with the decision, selected population, date window and required output. Record any constraint that could not be satisfied before presenting a conclusion. If the research changed scope, describe the change instead of making the final sample appear to answer the original brief. Keep the counting unit clear whenever a number appears.
Make claims traceable
For each material finding, retain original source links or authorized evidence references and explain what they support. Use small quotations where useful, keeping the author's context. Separate observed statements from your interpretation and from any independently verified product fact. Include counterexamples that materially weaken a proposed theme rather than selecting only convenient evidence.
Report coverage and uncertainty
State actual returned or inspected counts, freshness, source limits and unavailable records. A compact display is not necessarily the full dataset, and a full export is not proof of complete source access. Avoid invented market sizes, demographic claims or causal conclusions. When a number is derived from the inspected records, label its basis and calculation.
End with a bounded next step
Propose the smallest useful validation or action, such as checking a product behavior, interviewing a consenting participant or deliberately refreshing selected accounts. Keep recommendations distinct from executed effects. A brief should not secretly start recurring collection, export private records or contact people. Preserve the evidence reference so the same deliverable can be reviewed again without rerunning the research.
Keep the boundaries clear.
- A recommendation is not an executed action.
- Keep the evidence reference so a re-display does not become a new research run.
Start with a read-only check: ask your agent to report actual source coverage before collecting records or exporting an audience.
Primary references
Documentation reviewed 2026-11-13. Product capabilities, client availability and plans can change. Third-party marks identify their products, not an affiliation.