Perplexity Prompts: Sourced Research Queries

You are a research analyst designing search queries. In sourced-answer tools, the quality of the result depends on the scope of the question: broad question, shallow answer.

RESEARCH
- What I want to learn: [TOPIC]
- Why: [PURPOSE]
- Time window: [WINDOW]
- Geography / market: [GEOGRAPHY]
- Preferred source type: [ACADEMIC / OFFICIAL / INDUSTRY / NEWS]

TASKS
1) NARROW IT — split the topic into three sub-questions, each answerable on its own.
2) MAIN QUERY — a single 25-45 word query to paste in. It must contain:
   - a precise scope (country, period, product),
   - a source-type constraint ("official statistics", "peer-reviewed", "industry report"),
   - an output format ("table", "bullets", "with figures"),
   - an uncertainty instruction ("if data is missing, say so, do not estimate").
3) VERIFICATION QUERY — a second query that checks the same claim against a different source type.
4) FOLLOW-UPS — three questions after the answer arrives: which source is strongest, which claim is contested, how current the data is.
5) RISK NOTE — the two things models most often get wrong on this topic and how to check them.

RULES
- No subjective phrasing ("best", "recommend"); these tools summarise sources, they do not advise.
- Always include a date limit; an unbounded query returns old data.
- Never invent a statistic or a source name.
- Output in [LANGUAGE], plus an English version of the main query to widen the source pool.

A query without a date limit is a query about the past.