These are 24 complete, paste-ready prompts for doing real research with Perplexity — market and competitor analysis, literature reviews, fact-checking, trend scans, and cited summaries. Each one is written like a brief to an analyst: it names the source type to trust, sets a timeframe, demands a citation for every claim, tells the model to flag conflicting sources and list limitations, and specifies the output shape — a table, a brief, or a ranked list. Swap the bracketed placeholders and paste.
One habit runs through all of them: for anything deeper than a single fact, switch from a quick Search to Research mode, which runs a multi-step process, reads dozens of sources, and returns a cited report. Reach for Labs when you want the result assembled into a spreadsheet, dashboard, or deck. New to Perplexity? Start with the best Perplexity prompts roundup, and see how to prompt Perplexity for research for the method behind these.
Market & industry research
Five prompts for sizing and mapping a market. Turn on the Finance source toggle for market data, set a tight timeframe, and run these in Research mode so Perplexity reads across analyst reports, filings, and news rather than a single page.
1. Market Landscape Overview
Act as a market analyst. Give me a landscape overview of the [industry] market in [region], restricted to sources from the last 12 months. Cover: the main segments, the leading players and their rough share, the key demand drivers, and the biggest headwinds. Cite a source for every claim, prefer industry reports, official statistics, and reputable trade press, and flag any figure you could only find in a single source. Where estimates conflict between sources, show both and note which is more recent. Output as a structured brief with clear headings and a short "key uncertainties" list at the end.Best for: a fast, sourced orientation on an unfamiliar market — the "key uncertainties" list tells you where the data is thin before you rely on it.
2. Market Size & Growth Estimate
Estimate the current market size and forecast growth (CAGR) for [industry / product category] in [region]. Use only figures published in 2025 or 2026 and cite the source and publication date for each number. Present the estimates in a table with columns: metric, value, year, source, and methodology note. Because market-size figures vary widely by analyst, list at least three independent estimates where available, explain why they differ, and give a reasoned midpoint rather than a single headline number. End with the main assumptions and limitations behind these figures.Why it works: forcing three independent estimates into a table with dates and methodology exposes how much market-size numbers really disagree, instead of hiding it behind one confident figure.
3. Customer Segments & Needs
Map the main customer segments for [product / service] in [region]. For each segment, describe who they are, their primary jobs-to-be-done, their biggest unmet needs, and their typical budget or willingness to pay. Draw on industry research and, where relevant, real user discussion — turn on Social sources to capture actual complaints and desires from Reddit and forums. Cite every segment claim, and clearly separate evidenced findings from your own inference. Output as a table of segments plus a short paragraph on which segment looks most underserved and why.Best for: grounding a segmentation in real evidence — pairing industry sources with Social discussion surfaces needs that reports miss.
4. Regulatory & Risk Scan
Scan the regulatory and compliance landscape for [industry] in [region], covering only rules currently in force or announced within the last 18 months. For each relevant regulation, give its name, what it requires, who enforces it, key deadlines, and the penalty for non-compliance. Cite only official regulator publications, government sources, or primary legal texts — not secondary summaries — and link each one. Flag anything still in draft or under consultation separately from law that is already in effect. Output as a table sorted by deadline, followed by a short "biggest risks" summary.Why it works: restricting citations to primary regulator and legal texts keeps a compliance scan off blog paraphrases, and separating draft from enacted rules prevents acting on something that is not law yet.
5. Pricing Benchmark Across the Market
Benchmark the pricing of the leading [product category] offerings in [region] as of the last 6 months. For each named provider, capture their pricing tiers, what each tier includes, any usage limits, and free-trial terms. Cite the provider's own pricing page or official documentation as the source for every number, and note the date you would expect it to be current as of. Flag any price you could only find via a third party rather than the vendor directly. Output as a comparison table with one row per provider and a short note on which offers the best value at the [entry / mid / enterprise] tier.Best for: a defensible pricing comparison — insisting on the vendor's own page as the source keeps stale third-party numbers out of the table.
Competitor & company analysis
Four prompts for going deep on a specific company. Combine the Web and Finance toggles, and for public companies ask for official filings as the primary source. These are strong candidates for Labs when you want a sortable comparison spreadsheet.
6. Competitor Profile Deep Dive
Build a deep-dive profile of [company]. Cover: what they sell, their target customers, positioning and pricing, funding or ownership, approximate size (revenue and headcount if available), recent strategic moves in the last 12 months, and their apparent strengths and weaknesses. Cite a source for every factual claim, prefer official filings, the company's own site, and reputable news over unsourced blogs, and mark any estimate as an estimate. Flag where sources conflict. Output as a structured profile with headings, ending with a short "what this means for a competitor" takeaway.Why it works: asking it to mark estimates as estimates and to prefer filings over blogs keeps a competitor profile honest about what is confirmed versus guessed.
7. Head-to-Head Competitor Comparison
Compare [company A] and [company B] head-to-head for [use case / buyer type]. Build a comparison table with rows for: core product, target customer, pricing, standout strengths, notable weaknesses, integrations, and recent momentum (last 12 months). Cite a source for each cell and prefer primary sources — the companies' own materials and official filings — over review-site paraphrase. Where a claim rests on a single or low-quality source, mark it. After the table, give a two-paragraph verdict on which fits [use case] better and the main trade-off, with limitations noted.Best for: a buyer-facing comparison you can trust — the per-cell citations mean every row can be checked, and the flagged cells show where to look harder.
8. Company Financial Snapshot
Give me a financial snapshot of [public company], using only official filings, earnings releases, and investor materials as sources — turn on the Finance sources. Report the latest reported revenue, growth rate, gross margin, operating income or loss, cash position, and any guidance, each with the reporting period and a citation to the filing. Present it as a table of metric, value, period, and source. Note clearly where a figure is a company non-GAAP adjustment versus a standard measure, and list the key risks the company itself disclosed in its most recent filing.Why it works: tying every number to a filing and flagging non-GAAP adjustments stops a financial snapshot from blending marketing figures with audited ones.
9. Product & Feature Teardown
Do a feature teardown of [product] for [target user]. List its core capabilities, notable limitations, platform and integration support, and how users actually rate it — pull real sentiment from Social sources (Reddit, X, forums) alongside official documentation. For each capability, cite either the vendor's docs or a specific user report, and separate confirmed features from marketing claims you could not independently verify. Flag the most common complaint and the most praised strength. Output as a table of features plus a short "who it's for / who should avoid it" summary.Best for: a realistic product read — combining official docs with Social sentiment catches gaps between what a product claims and how it actually performs.
Academic & literature review
Four prompts for evidence-based questions. Turn on the Academic source toggle so Perplexity draws from published papers, insist on peer-reviewed citations, and use Research mode so it reads across the literature rather than one abstract.
10. Literature Review on a Research Question
Act as a research assistant and produce a mini literature review on the question: [research question]. Use Academic sources and cite only peer-reviewed studies, review articles, or meta-analyses, with the author, year, and a link for each. Organize the findings by theme rather than by paper, summarize what the weight of evidence says, and explicitly note where studies conflict or where the evidence is weak or preliminary. Distinguish correlational from causal findings. End with the main open questions and the limitations of the current literature.Why it works: organizing by theme and calling out conflicts, weak evidence, and correlation-versus-causation is what separates a real literature review from a pile of paper summaries.
11. State of the Evidence Summary
Summarize the current state of the evidence on whether [intervention / claim] works for [outcome]. Use only peer-reviewed studies and systematic reviews from the last 10 years via Academic sources, and cite each. Give me: the overall direction of the evidence, the strength and quality of that evidence, the size of any effect, key caveats (sample sizes, populations, funding conflicts), and where high-quality studies disagree. Be explicit about how confident the literature actually is. Output as a short brief ending with a one-line, appropriately hedged bottom line.Best for: answering a "does X work?" question responsibly — demanding the strength of evidence and a hedged bottom line prevents overstating a shaky result.
12. Find Foundational & Recent Papers
Help me build a reading list on [topic]. Using Academic sources, list (a) the foundational, most-cited papers that established the field, and (b) the most important papers from the last 2 years. For each, give the title, authors, year, venue, a one-sentence summary of its contribution, and a link. Prefer peer-reviewed work and note preprints separately. Organize the list from foundational to cutting-edge, and add a short note on which two or three papers to read first if I only have limited time.Why it works: splitting foundational from recent and flagging preprints gives a genuinely useful reading path rather than an undifferentiated citation dump.
13. Explain a Method or Concept with Sources
Explain [method / concept] clearly for someone with a [beginner / intermediate] background in [field]. Ground the explanation in Academic sources: define the concept, explain how and why it works, give a concrete example, and note the main limitations or common misconceptions. Cite a specific paper or authoritative text for each key claim, and distinguish established consensus from areas that are still debated. Keep it rigorous but readable. End with two or three primary sources I can read to go deeper.Best for: learning a concept without picking up errors — sourced explanations that separate consensus from debate are far safer than an unsourced summary.
Fact-checking & source-finding
Four prompts for verification. Keep Web on and lean on primary sources; the goal is to trace a claim back to where it started and see whether independent sources agree.
14. Verify a Specific Claim
Fact-check this claim: "[claim]". Search for the best available evidence and tell me whether it is supported, partly supported, disputed, or false. Cite the strongest primary and independent sources on each side, with dates. Trace the claim to its original source if you can, and note if it stems from a single origin that others merely repeat. Be explicit about what is established versus contested, and flag if the evidence is thin. Output a clear verdict line first, then the supporting and opposing evidence, then a confidence level and its limitations.Why it works: leading with a verdict then tracing the claim to its origin catches the common case where one shaky source gets repeated until it looks like consensus.
15. Find the Primary Source
I keep seeing this statement quoted: "[statement / statistic]". Find the original primary source it comes from — the specific study, dataset, filing, or official statement — not a news article repeating it. Give me the exact citation with author, date, and a direct link, and quote the relevant line from the original if possible. Note whether the way it is commonly quoted matches what the primary source actually says, and flag any distortion, missing context, or outdated figure. If you cannot locate a genuine primary source, say so plainly.Best for: chasing a viral stat back to the source — checking whether the popular phrasing matches the original often reveals the number has drifted.
16. Check a Statistic Across Sources
Verify this statistic: "[statistic]". Find every credible independent source that reports it or a related figure, published as recently as possible, and cite each with its date and methodology. Show whether the sources agree, and if the numbers differ, lay out the range and explain why — different definitions, timeframes, or samples. Identify the most authoritative and most current source. Output a short table of source, figure, date, and method, followed by a one-line assessment of what number is safe to cite and how to caveat it.Why it works: putting competing figures side by side with their definitions shows that most stat disputes are really definition disputes, and tells you which number to actually use.
17. Debunk or Confirm a Common Belief
Assess this widely held belief: "[belief]". Using credible, recent sources, tell me whether it is accurate, a myth, or an oversimplification. Cite the best evidence on each side, prefer primary research and authoritative bodies, and note where the belief was once true but no longer is, or is true only under specific conditions. Flag any source that seems biased or commercially motivated. Output a clear "accurate / partly true / myth" verdict, the reasoning with citations, and the main caveats — including how confident the evidence lets you be.Best for: pressure-testing conventional wisdom — the "true only under specific conditions" framing captures the many beliefs that are neither fully right nor fully wrong.
Trends & technology scans
Four prompts for spotting what is changing. Mix Web for reporting and Social for early signals, set a recent timeframe, and run in Research mode to catch developments across many sources at once.
18. Emerging Technology Scan
Scan for the most significant developments in [technology / field] over the last 6 months. Identify the notable advances, new entrants, funding rounds, and shifts in direction, each with a dated citation to a credible source. Separate confirmed developments from speculation and hype, and note where reporting relies on a single source or a company's own announcement. Prefer primary announcements, reputable technical press, and research over aggregator blogs. Output as a ranked list from most to least significant, with a short "why it matters" line for each and the limits of what is known.Why it works: separating confirmed advances from hype and ranking by significance turns a firehose of announcements into a signal you can act on.
19. Trend Signals from Social Sources
Using Social sources (Reddit, X, forums, community discussion), surface the emerging conversations and complaints around [topic / product category] in the last 3 months. Identify recurring themes, rising frustrations, unmet needs, and any shifts in sentiment, and link to representative threads or posts. Be clear that these are anecdotal signals, not statistics — quantify roughly how common each theme appears rather than overstating it. Flag anything that looks like coordinated promotion rather than organic opinion. Output as a ranked list of themes with a representative quote and link for each.Best for: catching early demand and pain signals — the reminder that these are anecdotal keeps you from treating forum sentiment as market data.
20. Timeline of Key Developments
Build a chronological timeline of the key developments in [topic / company / technology] from [start year] to now. For each entry, give the date, a one-line description of what happened and why it mattered, and a citation to a credible source. Prefer primary announcements and contemporaneous reporting over later retrospectives, and flag any date or event where sources disagree. Focus on genuinely pivotal moments rather than every minor update. Output as a clean dated timeline, ending with a short note on where the topic appears to be heading and how certain that read is.Why it works: preferring contemporaneous sources and flagging disputed dates keeps a timeline accurate where later retellings often blur or reorder events.
21. Who's Who in a Field
Give me a "who's who" of [field / topic]: the most influential people, organizations, and projects shaping it right now. For each, explain their role, their notable contribution, and why they matter, with a citation. Prefer primary sources and reputable coverage, and distinguish established, widely recognized figures from rising newer ones. Note any where influence is contested or self-claimed rather than independently recognized. Output as a grouped list (people, organizations, projects) with a one-line "known for" and a link for each entry.Best for: orienting fast in a new field — separating established names from rising ones tells you who to read now versus who to watch.
Summarize & synthesize sources
Three prompts for turning sources into something usable. These pair well with a Space: upload the documents, set standing instructions like "always cite" and "flag disagreements", and ask across all of them. Keep the Perplexity prompt cheat sheet open while you refine.
22. Summarize a Single Source with Citations
Summarize [this URL / uploaded document] accurately and concisely. Give me: the main argument or finding, the key supporting points, any data or evidence cited, the stated limitations, and the author's apparent perspective or potential bias. Quote the exact lines behind any important claim so I can verify them, and do not add facts that are not in the source. If the source makes a claim without support, say so. End with three sentences on what the source does and does not establish, staying strictly within what it actually says.Why it works: quoting the exact lines and forbidding outside facts keeps a summary faithful to the source instead of quietly importing the model's own assumptions.
23. Synthesize Multiple Sources into a Brief
Synthesize the following sources on [topic] into a single decision-ready brief: [list URLs or uploaded files]. Pull out the points where the sources agree, where they disagree, and what each adds uniquely, citing which source supports each statement. Do not blend conflicting claims into a false consensus — surface the disagreements explicitly. Weight more credible and more recent sources higher and say when you are doing so. Output as a brief with sections for consensus, open disagreements, and implications, ending with the strongest-supported takeaway and its limitations.Best for: turning a stack of reading into a usable brief — refusing to smooth over disagreements preserves the tension that actually matters for a decision.
24. Compare What Sources Disagree On
Compare how these sources treat [question / topic]: [list URLs or uploaded files]. For each point of contention, lay out each source's position side by side, cite the exact claim, and explain why they differ — different data, timeframes, definitions, methods, or incentives. Identify which position has the stronger evidence and which is weaker, and note where the disagreement cannot yet be resolved from available evidence. Output as a table with one row per disputed point (columns: the question, each source's view, who's better supported) plus a short verdict on where the truth most likely sits.Why it works: a side-by-side table of positions with a reasoned verdict makes disagreement legible and points you to the evidence that would settle it.
Once these become second nature, build your own from the same parts — source type, timeframe, citations, flag conflicts, output shape. Keep the cheat sheet handy, browse the full best Perplexity prompts roundup, and read how to prompt Perplexity for research for the reasoning behind each line.
Frequently Asked Questions
When should I use Perplexity's Research mode instead of a quick Search?
Use a quick Search for a single fact or a fast answer you can verify at a glance. Switch to Research mode for any real deep dive: it runs a multi-step process, reads dozens of sources, and returns a structured, cited report instead of a short answer. If your prompt asks for a comparison table, a literature review, a market landscape, or anything where you need conflicting sources reconciled, choose Research. It is slower but far more thorough, and every claim comes back with a citation you can click through to check.
How do I get Perplexity to cite only credible sources?
Say so explicitly in the prompt and name the source type you will accept: "cite only peer-reviewed studies", "use only official company filings and primary sources", or "government statistics agencies and regulator publications only." Then turn on the matching source toggle — Academic for papers, Finance for market data — so Perplexity draws from the right pool. Ask it to give a citation for every claim and to flag any statement it could only support with a low-quality or single source, so you know where to double-check.
What are the Web, Academic, Social, and Finance source toggles for?
Perplexity lets you steer which pool it searches. Web is the general default. Academic restricts results to published papers and is the one to use for literature reviews and evidence-based questions. Social pulls opinions and discussion from Reddit and X, which is useful for sentiment, real-world complaints, and early signals. Finance focuses on market data, filings, and earnings. Match the toggle to the job: a fact-check leans on Web plus primary sources, a lit review on Academic, a product-sentiment scan on Social.
What is Perplexity Labs and when should I use it?
Labs goes a step beyond Research: instead of returning a report, it builds a deliverable — a spreadsheet, a dashboard, a mini web app, or a slide deck — from your research. Use it when the output you actually need is an artifact rather than prose: a competitor-comparison spreadsheet you can sort and filter, a tracked-metrics dashboard, or a first-draft deck. It runs the same multi-source research underneath, then assembles the result into the format you asked for.
What makes a research prompt on Perplexity actually good?
Five things: name the source type you will accept, set a timeframe like "the last 12 months", demand a citation for every claim, tell it to flag conflicting sources and list limitations, and specify the exact output shape — a comparison table, a brief, a ranked list. Vague prompts get vague answers; a good research prompt reads like a brief to an analyst, with the standard of evidence and the deliverable spelled out. Every prompt on this page is built that way, so you can lift it and swap the bracketed details.
How do Spaces help with a long research project?
A Space is a persistent workspace where you can upload your own files — PDFs, reports, spreadsheets — and set custom instructions that apply to every question you ask inside it. For a project that runs over days or weeks, a Space keeps Perplexity grounded in your documents and your standing rules ("always cite sources", "prefer primary data", "focus on the EU market") without repeating them each time. It is the right home for a literature review, a due-diligence file, or an ongoing market watch.
Why should I set a timeframe in a research prompt?
Because research questions decay. Without a window, Perplexity may mix a five-year-old statistic with last month's news and present both as current. Adding "restrict to the last 12 months" or "published in 2025 or 2026" forces it to weight recent, relevant sources and to flag when the freshest data it can find is older than your window. For fast-moving topics — pricing, regulation, technology — the timeframe is often the single most important line in the prompt.
Can I trust Perplexity's citations without checking them?
Treat citations as a starting point, not a guarantee. Perplexity links the sources it used, which is a large improvement over a chatbot that invents references, but a linked source can still be misread, outdated, or itself unreliable. For anything that matters — a decision, a published claim, a number you will quote — click through and confirm the source says what the summary claims. Asking the model to flag single-source or conflicting claims tells you exactly which lines deserve that extra check.