The prompt
Score 10 of 10, Excellent
Use it when Someone asked you for "the numbers on X" and you need to pin down the question, the decision, the definitions and the deadline before pulling any data.
- {{requester}}
- {{request}}
- {{data_available}}
- {{constraints}}
Full prompt and why it works
Prompt
# Role You help analysts turn loosely worded data requests into a brief that both sides agree on before any work starts. # Goal Turn the request below into a one-page analysis brief that I can send back to {{requester}} to confirm, so we agree on the question, the definitions and what will be delivered before I spend time pulling data. # Success criteria - The brief states the decision the analysis will inform, not only the numbers asked for. - Every metric has a definition precise enough to calculate, including the period, filters and what is excluded. - The requester could reply "yes" or correct one line, without needing a meeting. # Output Use these headings: 1. The question, in one sentence. 2. The decision it informs, and who makes it. 3. Metrics and definitions: a table with the columns Metric, Definition, Source table or system, Period, Filters. 4. What will be delivered: the format (table, chart, short memo or dashboard), the level of detail and the delivery date. 5. Out of scope: what this request will not cover. 6. Assumptions, numbered, for the requester to confirm. 7. Questions for the requester: up to 5, most important first. 8. A cover message under 80 words asking them to confirm or correct the brief. For example, a metric definition might read: "Active customer: placed at least one paid order in the 90 days before the end of each month; excludes staff and test accounts." This example shows the form only. # Constraints - Use only what the request and my notes say. Where they leave something open, write it as an assumption or a question, never as a settled fact. - Do not start the analysis or suggest what the answer will be. - Keep the brief to one page, under 450 words, not counting the cover message. # Stop rules If the request does not say what decision it supports, put that first in the questions instead of guessing one, because the decision decides which numbers matter. # Context The request as it was sent: """ {{request}} """ What I know about the data available (tables, systems, known gaps): """ {{data_available}} """ My deadline and other work competing for time: {{constraints}}
Variables
- {{requester}}
- Who asked, for example "the head of marketing".
- {{request}}
- The request exactly as it reached you: the email, chat message or ticket.
- {{data_available}}
- The tables or systems that hold the relevant data and any gaps you know of.
- {{constraints}}
- When you could deliver and what else is competing for your time, for example "other priorities until Wednesday; could deliver Friday".
Why it works
- Who it's for. The brief starts from the decision the numbers will inform, which is what tells you which numbers are worth pulling.
- What good looks like. Each metric needs a definition precise enough to calculate, so "active customers" means the same thing to both sides.
- Layout. Markdown sections for role, goal, success criteria, output, constraints and stop rules, with the pasted request last, the layout OpenAI recommends.