These are 35 complete, paste-ready prompts for Perplexity, the AI answer engine that returns cited answers with numbered inline citations. Each one names the source type, scope, timeframe, and output format so you get a verifiable brief instead of a vague reply, and each notes which mode — quick Search, Research, or Labs — and which Focus toggle to use. Drop any of them straight into Perplexity and adjust the bracketed details.

Want to go deeper on one use? Read how to prompt Perplexity for research, keep the Perplexity prompt cheat sheet open while you write, and browse the dedicated research prompts collection when you need more depth.

Advertisement

Research & deep dives

Six prompts for multi-source reports. Run these in Research mode so Perplexity plans, reads dozens of sources, and returns a structured cited report — switch the Focus toggle to Academic when the question is evidence-heavy.

1. Structured Research Brief on a Topic

Research [topic] and return a structured brief. Cover: (1) what it is and why it matters, (2) the current state as of the last 12 months, (3) the main competing viewpoints, (4) the strongest supporting evidence, and (5) open questions. Use primary sources and reputable reporting; prefer sources from the last 12 months. Link every claim to a numbered citation. End with a "Confidence & limitations" section that flags where sources conflict or evidence is thin. Format as a bulleted brief with clear headings.

Why it works: It gives Perplexity a fixed report skeleton, a timeframe, a source rule, and an explicit uncertainty section — the four things that turn a Research run into a brief you can act on.

2. State of the Field Literature Scan

Give me a state-of-the-field scan on [research area]. Cite only peer-reviewed studies, systematic reviews, or official reports — no blogs or press releases. For each key finding, note the study, year, sample or scope, and how strong the evidence is. Group findings into "well-established", "emerging", and "contested". Prefer work from the last 5 years but include landmark older studies. Flag any conflicting results explicitly. Present as a table: Finding | Evidence strength | Key source.

Best for: Grounding a paper or decision in real evidence — pair it with the Academic Focus toggle so it only reads scholarly sources.

3. Two-Sided Debate Map

Map the debate around [contested question]. Present the strongest case for each side fairly, using the best sources each side actually relies on. For every argument, give the core claim, the evidence behind it, and the main rebuttal. Do not take a side. Prefer primary sources and named experts over opinion pieces. End with the key points where both sides agree and the specific facts that are still genuinely disputed. Link every claim. Format as two columns plus a shared "Points of agreement" section.

Why it works: Asking for the best sources each side relies on, plus a shared-agreement section, produces a balanced map instead of a one-sided summary dressed up as neutral.

4. Competitive Landscape Report

Research the competitive landscape for [industry or product category]. Identify the top 6-8 players, and for each give: positioning, target customer, pricing model, notable strengths, and a known weakness or criticism. Use company filings, official pricing pages, and recent reporting from the last 12 months. Note where public information is unavailable rather than guessing. Link every data point to a source. Return as a comparison table, then a 3-sentence takeaway on where the market is heading.

Best for: Market and strategy prep — the "note where information is unavailable rather than guessing" line stops the model from inventing pricing or numbers.

5. Timeline of How a Situation Developed

Build a dated timeline of how [event or situation] developed, from the earliest relevant moment to today. For each entry give the date, what happened, and a citation to a primary or contemporaneous source. Distinguish confirmed facts from later claims or allegations. Prefer official statements, court or regulatory filings, and reporting from the time. Flag any dates or facts that sources disagree on. Format as a chronological list with a citation on every entry.

Why it works: Forcing contemporaneous primary sources and separating confirmed facts from allegations keeps the timeline accurate rather than a retelling of the latest hot take.

6. Deep Explainer for a Non-Expert

Explain [complex topic] to a smart non-expert in about 600 words. Start with a one-paragraph plain-English summary, then build up the key mechanisms and why they matter. Use a concrete analogy where it helps, define any jargon on first use, and cite a reliable source for every non-obvious claim. Prefer authoritative primary sources. End with 3 things people commonly get wrong about it, each with a citation. Keep it accurate over simple — if something is genuinely uncertain, say so.

Best for: Getting up to speed fast — the "accurate over simple" instruction and the common-misconceptions list stop it from smoothing over the hard parts.

Fact-checking & sources

Five prompts for verification. These lean on primary sources and cross-checking; use quick Search for a single claim and Research when a story has many moving parts. For more, see the research prompts collection.

7. Verify a Specific Claim

Fact-check this exact claim: "[paste the claim]". Rate it as True, Mostly true, Mixed, Mostly false, or False, and explain why in 3-4 sentences. Base the verdict only on primary sources or reputable fact-checkers, and quote the specific line from a source that supports your rating. If the claim is partly true, separate the accurate part from the inaccurate part. If it can't be verified, say "Unverifiable" and explain what evidence is missing. Link every source.

Why it works: A fixed rating scale plus a required supporting quote forces the model to point at evidence rather than assert a verdict, and the "Unverifiable" option prevents false confidence.

8. Trace a Statistic to Its Primary Source

Find the original primary source for this statistic: "[paste the stat]". Trace it back past any blogs or news articles that merely repeat it, to the study, dataset, or official report where it first appeared. Tell me who produced it, the year, the exact figure and its wording, the sample or methodology, and any caveats the source itself gives. Note if the number is commonly quoted out of context. Link the primary source directly.

Best for: Before you cite a number — chasing it past the repeaters to the origin often reveals a caveat everyone dropped.

9. Check a Viral or News Story

I saw this story circulating: "[paste headline or summary]". Check what actually happened using reporting from the last 7 days and any official statements. Separate confirmed facts from speculation and from what is still unconfirmed. Note which outlets are reporting it and whether any are contradicting each other. If it's misleading or missing context, explain how. Prefer primary sources over aggregators. Give a short verdict, then the evidence, with every claim linked.

Why it works: The 7-day window and the demand to flag contradicting outlets catch the churn of a developing story instead of freezing an early, wrong version of it.

10. Cross-Check Conflicting Sources

There's disagreement about [question]. Find the range of positions across reputable sources and lay out who claims what and on what basis. Where sources conflict, tell me why — different data, different timeframes, different definitions, or bias. Identify which position the strongest primary evidence supports, and be explicit about how confident that conclusion is. Do not paper over the disagreement. Link every source and format as: Position | Who holds it | Evidence | Reliability.

Best for: Messy topics where a single answer would be dishonest — the table makes the disagreement itself the deliverable.

11. Vet a Company or Person's Reputation

Give me a balanced reputation check on [company or person]. Cover verified facts, notable achievements, and any controversies, lawsuits, or credible criticism — each with a citation. Use official records, court or regulatory filings, and reputable reporting; check Reddit and X for first-hand experiences but label those as anecdotal. Separate confirmed issues from rumors. Flag anything you can't verify. Prefer sources from the last 24 months for current standing. End with a one-line, evidence-based summary.

Why it works: Mixing official filings with clearly labeled social anecdotes gives you both the record and the lived experience without confusing the two — set Focus to Social for the second half.

Shopping & buying decisions

Five prompts to buy smarter. Set the Focus toggle to Social to pull real owner opinions from Reddit and X, and ask for current pricing so the answer reflects today, not last year.

12. Head-to-Head Product Comparison

Compare [Product A vs Product B] for someone who will mainly use it for [use case]. Build a table covering price, key specs, real-world performance, reliability, and the biggest complaint owners report about each. Pull current prices and specs from official pages and recent hands-on reviews from the last 12 months; pull complaints from Reddit and owner forums. Say which one wins for my use case and why, and name the one situation where the other is the better pick. Link every source.

Best for: The two-way decision — anchoring the comparison to a specific use case forces a real recommendation instead of a spec dump.

13. Best Option Under a Budget

Find the best [product type] I can buy for [your budget] in [country/region], for [use case]. Give a top pick plus two alternatives, each with current price, why it fits my budget and needs, and its main trade-off. Use current pricing and reviews from the last 12 months; check owner discussions on Reddit for reliability red flags. Exclude anything discontinued or hard to buy where I am. Flag if my budget is unrealistic for what I want. Link prices and reviews.

Why it works: Budget, region, and use case in one line lets Perplexity filter to what you can actually buy, and asking it to flag an unrealistic budget saves a wasted search.

14. Real-Owner Opinions from Social

What do real owners actually think of [product] after living with it? Focus on Reddit, forums, and X rather than sponsored reviews. Summarize the most common praise, the most common complaints, and any recurring reliability or long-term issues. Note roughly how widespread each opinion is and how recent, and quote a representative comment for the big themes with a link. Call out anything that looks like astroturfing or paid hype. Be honest if sentiment is mixed.

Best for: Cutting past marketing — the Social Focus toggle plus a request for representative quotes surfaces the complaints reviews leave out.

15. Is It Worth Upgrading?

I currently have [current model/version] and I'm considering [newer model/version]. Is the upgrade worth it for someone who mainly does [use case]? List the meaningful differences, which ones I'd actually notice, and which are marketing. Use official specs plus hands-on comparisons and owner reports from the last 12 months. Give a clear verdict — upgrade, wait, or skip — with the price of the upgrade weighed against the real benefit. Flag any known issues with the new model. Link every claim.

Why it works: Separating differences you'd notice from marketing, and pricing the upgrade against real benefit, is exactly the judgment a spec sheet won't make for you.

16. Find the Catch Before You Buy

I'm about to buy [product or service]. Tell me the downsides before I commit. Find the most common complaints, hidden costs, subscription traps, poor support experiences, or reliability problems people report. Use owner discussions, review sites, and complaint threads from the last 12 months. Note how common each issue is and whether the company has addressed it. Include the return or cancellation policy if relevant. Be blunt — I want the reasons not to buy, each with a source.

Best for: A final gut-check — explicitly asking for reasons not to buy counters the model's tendency to be agreeable.

Advertisement

Work & productivity

Five prompts for the workday. A couple of these are Labs prompts that build a real deliverable; save recurring ones to a Space with your format and source rules baked in.

17. Meeting Prep Briefing

I have a meeting with [company or person] about [topic]. Prepare a one-page briefing. Cover: who they are and what they do, recent news or announcements from the last 6 months, likely priorities or pain points, and 5 smart questions I could ask. Use official sources, recent reporting, and their own public statements. Note anything I should avoid assuming. Flag where information is thin. Keep it to bullets I can skim in two minutes, with a citation on every factual claim.

Why it works: The 6-month window keeps the briefing current, and requiring a citation per fact means you walk in with claims you can stand behind.

18. Summarize and Extract Actions from a Document

Here is a document: [paste text or upload the file to this Space]. Give me a tight summary in 5 bullets, then a separate list of every action item, decision, deadline, and owner mentioned. Quote the exact line each action comes from so I can verify it. Flag anything ambiguous or where an owner or date is missing. Do not add anything that isn't in the document. End with the 3 most important open questions the document leaves unanswered.

Best for: Long reports and transcripts — uploading the file to a Space keeps the summary grounded in your document, and the quote-the-line rule makes every action checkable.

19. Draft a Policy from Best Practice

Draft a [type of policy, e.g. remote-work or data-retention] policy for a [size and type] organization in [country/region]. Base it on current best practice and any legal requirements that apply, citing the specific regulation or standard for each requirement. Keep it clear and practical, with sections for scope, rules, responsibilities, and review. Flag anything that genuinely needs a lawyer's sign-off rather than guessing. Note where requirements differ by jurisdiction. Link every legal or standards citation.

Why it works: Tying each requirement to a named regulation and flagging what needs a lawyer keeps a useful draft honest about its limits.

20. Build a Comparison Spreadsheet in Labs

Using Labs, build a comparison spreadsheet of [items to compare, e.g. project-management tools] for a [team size and use case]. Columns: name, price per user, key features, integrations, notable limitation, and a source link. Fill it from official pricing pages and recent reviews from the last 12 months, and leave a cell blank rather than guessing when data is unavailable. Add a short recommendation row at the bottom for my use case. Make the sheet downloadable.

Best for: Vendor selection — Labs turns the research into a spreadsheet you can hand to a team, with a source link in every row.

21. Turn Research into a Slide Deck

Using Labs, research [topic] and turn it into a 10-slide deck for [audience]. Structure: title, the problem, why now, 3-4 key findings each backed by a cited source, implications, and recommended next steps. Keep each slide to a headline and 3-4 bullets, with the source noted on any slide that states a fact or number. Prefer sources from the last 12 months. Add a final "Sources" slide listing every citation. Make the deck downloadable.

Why it works: Labs handles both the research and the artifact in one pass, and forcing a source on every fact slide keeps a good-looking deck defensible.

Marketing & content

Five prompts for content and growth. Ask for current data and real audience language; set Focus to Social when you want the words your customers actually use.

22. Keyword and Question Research

Research what people actually search and ask about [topic/niche]. Group the questions and phrasings by intent — informational, comparison, and buying — and note the real language people use. Pull from forums, Reddit, Q&A sites, and recent articles rather than guessing. For each cluster, suggest one content angle that would genuinely help. Flag questions that seem to have poor existing answers as opportunities. Link representative sources for each cluster. Return as a grouped list.

Best for: Planning a content calendar — grouping by intent and flagging under-served questions points you at the topics worth writing.

23. Competitor Content Audit

Audit how [competitor] approaches content for [topic/niche]. Identify the main themes they cover, their apparent angle and tone, formats they use, and any obvious gaps they're not addressing. Use their own published pages and recent reporting from the last 12 months. Don't guess at private metrics — stick to what's publicly visible and say so. End with 5 specific opportunities where I could do better or cover something they miss. Link every example page.

Why it works: Restricting the audit to publicly visible pages and forbidding guessed metrics keeps it grounded, and the gap list turns observation into a plan.

24. Trend Brief with Sources

Give me a trend brief on what's changing in [industry/topic] right now. Cover the 5-6 most significant recent developments from the last 6 months, why each matters, and who's driving it. Distinguish genuine shifts from hype, and cite a credible source for each. Note anything being over-claimed. Prefer primary sources, official announcements, and reporting over opinion pieces. End with what this means for [my role/business] in 3 bullets. Link every development.

Best for: Staying current without drowning — asking it to separate genuine shifts from hype filters the noise the news cycle amplifies.

25. Audience Pain-Point Mining

Find the real frustrations and unmet needs of [target audience] around [problem/product category]. Mine Reddit, forums, reviews, and X for the exact words people use to describe their pain. Group the pain points by theme, rank them by how often they come up, and quote a representative comment for each with a link. Note which frustrations existing products fail to solve. Don't sanitize the language — I want how customers actually talk. Flag anything that looks like an outlier.

Why it works: The Social toggle plus a demand for verbatim language hands you copy angles and product ideas in your customers' own words, backed by linked comments.

26. Fact-Check a Draft Before Publishing

Here is a draft article: [paste text]. Check every factual claim, statistic, date, and name in it. For each, tell me whether it's correct, questionable, or unverifiable, and cite a primary source. List any claims that are outdated, missing context, or need a source added. Do not rewrite the piece — just give me a checklist of issues to fix, ordered by risk. Flag the single claim most likely to draw a correction. Link every source you used to verify.

Best for: The pre-publish pass — a risk-ordered checklist catches the claim most likely to get you a correction before readers do.

Advertisement

Finance & investing

Four prompts for money decisions. Set the Focus toggle to Finance so Perplexity targets tickers, analyst ratings, and SEC filings — and always open the cited filings before you act.

27. Company Fundamentals Snapshot

Give me a fundamentals snapshot of [ticker]. Cover: latest revenue and earnings trend, margins, debt, cash position, valuation multiples versus peers, and the main growth drivers and risks. Pull figures from the most recent official filings and earnings reports, and date every number. Cite the filing for each figure. Distinguish reported facts from analyst estimates. Flag anything unusual or one-off. This is not financial advice — end with the 3 things I'd need to verify before making a decision. Format as a table plus a short summary.

Why it works: Dating every number, citing the filing, and separating reported facts from estimates is how you avoid acting on a stale or estimated figure — the Finance toggle keeps sources on-target.

28. Analyst Ratings and Sentiment Roundup

Summarize current analyst ratings and sentiment on [ticker]. Give the range of ratings and price targets, who's bullish and who's bearish and why, and how sentiment has shifted over the last 3 months. Separately, note what retail investors are saying on Reddit and X, labeled as sentiment not analysis. Cite each rating and claim, and date it. Flag where analysts sharply disagree. Remind me this is not advice. Format as: Bull case | Bear case | Retail sentiment, each with sources.

Best for: Seeing both sides — splitting professional ratings from clearly labeled retail sentiment keeps the crowd's mood from masquerading as analysis.

29. Read the Latest Earnings Filing

Read [ticker]'s most recent earnings report and filing and tell me what actually matters. Cover: did they beat or miss on revenue and earnings, what guidance did they give, what changed versus last quarter, and what management flagged as risks. Quote the specific numbers and the exact guidance language, citing the filing. Separate the headline from the details that the market may have overlooked. Note anything buried in the footnotes. This is not advice. Keep it to a one-page brief with every figure sourced.

Why it works: Quoting the exact guidance language and pulling from the footnotes surfaces the detail behind the headline beat-or-miss that moves a stock.

30. Compare Two Investments

Compare [investment A vs investment B, e.g. two ETFs or two stocks] for a [time horizon and risk tolerance] investor. Build a table of the factors that matter — for funds that's expense ratio, holdings, historical return, and risk; for stocks that's valuation, growth, and balance sheet. Use official fund documents or filings and date every figure. Explain the key trade-off between them in plain English. This is not financial advice — flag what depends on my personal situation. Cite every data point.

Best for: A structured A-vs-B — pulling from official fund documents and flagging what depends on your situation keeps a comparison from reading like a recommendation it can't responsibly make.

Learning & study

Five prompts for learning anything. Use the Academic Focus toggle for evidence-based subjects, and ask for citations so you can check the source rather than trust a summary.

31. Explain a Concept at Three Levels

Explain [concept] at three levels: (1) to a curious 12-year-old, (2) to a college student, and (3) to a professional in the field. Each level should be accurate, not just simpler — build real understanding, not a watered-down version. Define key terms, use one good analogy per level, and cite a reliable source for the core facts. End with a short "common misconceptions" note. If experts genuinely disagree on any part, say so instead of picking one view.

Why it works: Three calibrated levels plus a misconceptions note lets you enter at your depth and still catch the errors a plain summary would hide.

32. Build a Study Plan from Sources

Create a [number]-week study plan to learn [subject] from scratch to [goal level]. Break it into weekly milestones with specific topics, and for each recommend the best free or reputable resource — courses, papers, docs, or books — with a link. Order topics so each builds on the last. Include checkpoints to test myself. Prefer authoritative, current sources and note if any recommendation costs money. Flag topics that are commonly skipped but important. Return as a week-by-week table.

Best for: Self-teaching — a sequenced plan tied to real, linked resources beats a generic outline you'd still have to go source yourself.

33. Generate a Practice Quiz with Citations

Make a 10-question practice quiz on [topic] at [difficulty level]. Mix multiple-choice and short-answer, covering the most important ideas. After the questions, give an answer key that explains why each answer is correct and cites a reliable source for the key facts. Make the wrong options plausible, not obvious. Flag any question where the "correct" answer is genuinely debated among experts, and explain the debate. Keep questions and answers clearly separated so I can test myself first.

Why it works: Citing the answer key and flagging genuinely debated questions turns a quiz into a study tool you can trust, not just a memory drill.

34. Summarize a Paper for a Beginner

Summarize this paper for someone new to the field: [paper title, link, or upload]. Cover the research question, what they did, what they found, and why it matters, in plain language. Define technical terms on first use. Separately note the study's limitations and how strong the evidence actually is — sample size, method, whether it's been replicated. Don't overstate the findings. If the paper's claims are contested or preliminary, say so. Cite the specific sections you're drawing from.

Best for: Reading outside your field — the limitations-and-strength section stops a single flashy paper from being taken as settled science.

35. Compare Competing Theories or Methods

Compare the leading theories or methods for [topic, e.g. approaches to X]. For each, explain the core idea, its main strengths, its known weaknesses, and the kind of evidence that supports it. Note where the field currently leans and where it's genuinely unsettled. Cite peer-reviewed or authoritative sources, prefer recent work, and flag any that are outdated or superseded. Don't force a single winner if the evidence doesn't support one. Format as a table plus a short synthesis.

Why it works: Asking it not to force a winner when evidence is split — and to flag superseded approaches — gives you the honest lay of the land instead of a false consensus.

Once a prompt lands close, refine it the way you'd brief a researcher: tighten the scope, name the exact source type, and ask it to flag uncertainty. Read how to prompt Perplexity for research for the full method, keep the cheat sheet open while you write, and browse the research prompts collection when a question needs a full report.

Frequently Asked Questions

Which Perplexity mode should I use — Search, Research, or Labs?

Use quick Search for a single fact or a fast cited answer. Switch to Research mode when the question needs multiple steps and dozens of sources — it plans, reads widely, and returns a structured cited report, which is what most prompts on this page are written for. Use Labs when you want a finished deliverable rather than an answer: a spreadsheet, a dashboard, a slide deck, or a small web app. A good rule of thumb is Search for facts, Research for reports, and Labs for artifacts.

How do I get Perplexity to cite better sources?

State the source type inside the prompt. Ask it to cite only peer-reviewed studies, official filings, or primary documents, to prefer sources from the last 12 months, and to link every claim to a numbered citation. Then set the Focus toggle to match: Academic for papers, Finance for tickers and filings, Social for real user opinions from Reddit and X. Telling the model both the source type and the timeframe is the single biggest lever on citation quality.

What are the Focus or Source toggles for?

The Focus toggles narrow where Perplexity searches. Web is the general default. Academic restricts to scholarly papers, which is best for research and evidence questions. Social pulls from Reddit and X, which surfaces honest first-hand opinions for shopping and reputation checks. Finance targets tickers, analyst ratings, and SEC filings for investing work. Matching the toggle to your question type removes noise before the model even starts reading.

Which model should I pick in Perplexity Pro or Max?

Sonar Pro is the fast default and handles most cited answers well. Switch to Sonar Reasoning Pro when the question needs multi-step logic, comparison, or synthesis. Pro and Max users can also select frontier models — GPT-5.2, Claude Opus 4.6, or Gemini 3 Pro — for the hardest reasoning or writing. Max adds Model Council, which queries three frontier models and synthesizes their answers, which is worth it for high-stakes decisions where you want cross-checking built in.

Why should I ask Perplexity to flag uncertainty?

Answer engines sound confident even when sources disagree, so asking the model to flag conflicting sources, note where evidence is thin, and list its limitations turns a tidy answer into an honest one. Every prompt on this page that touches a decision ends with an instruction to surface disagreement and uncertainty. It is the difference between a summary you can act on and one you have to re-verify from scratch.

What are Spaces and when should I use one?

Spaces are persistent workspaces where you can save custom instructions and upload your own files. Use one when you run the same kind of query repeatedly — set your preferred output format, source rules, and tone once, and every thread in that Space inherits them. Upload a report, a dataset, or a set of PDFs and Perplexity will answer against your documents alongside the web, which keeps recurring research consistent.

Can I trust Perplexity's answers without checking the sources?

Treat the answer as a well-sourced first draft, not a final one. The numbered citations exist so you can verify each claim, and you should open the load-bearing ones before you act, especially for finance, health, or legal questions. Prompts that force primary sources and flag uncertainty make this faster, but the model can still misread a source or cite a weak one, so the citations are there to be clicked, not just counted.

What makes a great Perplexity prompt different from a Google search?

A search box wants keywords; Perplexity wants a brief. The best prompts state the scope, the timeframe, the source type you trust, and the exact output format — usually a table or a bulleted brief with citations. Because it returns a synthesized cited answer instead of a list of links, spelling out those four things up front is what turns a vague reply into a precise, verifiable one you can paste into your work.

Advertisement