Perplexity Citations: How to Get Cited and Track Where You Appear

TL;DR. Perplexity answers every question by searching the live web, then attaching some of the pages it read to specific claims in the answer. Those links are its citations. Being listed as a source, being cited next to a claim, and being named as the recommendation are three different outcomes, so track them separately. Pages get cited when PerplexityBot can reach them, when they answer the exact question near the top, and when they carry facts Perplexity can check: license numbers, prices, dates, named authors. To track citations, run a fixed set of buyer questions on a schedule and log all three outcomes, or automate it through Perplexity's API for well under a dollar a month.

Product and API facts on this page were checked against Perplexity's Help Center and API documentation on September 21, 2026, and the example answers were run in Perplexity the same day. Perplexity ships changes often, so the sources at the bottom are the ones to recheck.

What are Perplexity citations?

A Perplexity citation is a link from a sentence in an answer to the web page that supports it. Perplexity searches the web for each question, writes an answer, and attaches sources to the claims so readers can check them. Its help center describes every answer as carrying numbered citations back to the original sources. Perplexity reports 1.5 billion questions a month, so those links are a real distribution channel for the sites that earn them.

Here is where a citation shows up in the web app today:

Perplexity's documentation still describes citations as numbered, and its API still marks them [1], [2]. The web app shows the same links as named chips. You can also highlight any passage in an answer to see the sources behind that passage alone.

The detail most guides miss: the snippet in the hover card and the Links tab is often your meta description, word for word. In our test, the cited pages showed their meta descriptions verbatim, under a shortened or lightly edited version of the title tag. Your title and meta description are the preview a Perplexity user reads before deciding whether to click.

Retrieved, cited, recommended: three different outcomes

"Did Perplexity cite us?" hides three separate questions:

1. Retrieved. Your page is listed in the Sources panel or the Links tab. Perplexity read it, but may not have used it.
2. Cited. Your page is attached to a specific claim in the answer. This is the citation that earns clicks.
3. Recommended. Your business or product is named as the answer ("start with X"). On "best plumber near me" questions, this is the outcome that produces calls, and it can happen even when your own site isn't cited, because the facts came from Yelp or a directory.

Perplexity's developer docs draw the same line: their citation-parsing example reports how many sources a response found and, separately, how many it cited. In our emergency-plumber test below, the answer read 29 sources, cited a smaller subset of them inline, and recommended two businesses. The citation vs recommendation glossary entry covers the distinction in more depth.

How does Perplexity choose which sources to cite?

Perplexity doesn't publish a ranking formula. Here is what it does document, and what shows up consistently in testing.

What Perplexity documents

• It searches live, from its own index. Perplexity's crawler, PerplexityBot, exists to find sites and link them in Perplexity's search results. Perplexity says it is not used to crawl content for training AI foundation models.
• It fetches pages on demand. A second agent, Perplexity-User, visits pages while Perplexity answers a user's question. Because a person triggered the fetch, Perplexity says this agent generally ignores robots.txt.
• It keeps two dates for every source. The API returns a publication date and a last_updated date for each result, and developers can filter results by either. Freshness is part of the data Perplexity stores about your page.
• It labels some sites. Perplexity's source review rates whole websites, not single pages, and gives some a Government, Academic, or Trusted label. The review asks objective questions, including whether the site corrects its mistakes, says who wrote each piece, and keeps news separate from advertising and opinion. Perplexity says partnerships and payments don't affect labels. Most domains have no label, and Perplexity says that is not a negative judgment.
• Deeper modes read more. Perplexity's help center says its research mode runs dozens of searches and reads hundreds of sources. A standard answer reads fewer, which leaves fewer citation slots.

What it doesn't publish, and what testing shows

These are observations, not documented rules:

• Exact-question relevance beats general authority. Pages that answer the precise question get cited over bigger sites that cover the topic broadly.
• Checkable specifics get quoted. Numbers, names, policies, and dates end up in answers. Adjectives don't.
• Agreement matters. When several independent sites state the same fact, answers state it plainly and cite more than one of them. When sources disagree, answers hedge, for example "review totals vary by directory."
• The source mix follows the question. For "perplexity citation tracking," the sources were two Perplexity developer docs, six tracking-tool product pages, two "best tools" roundups, and a Reddit thread. A local "who's best" question pulls a completely different set, as the next example shows.

A real example: "who is the best emergency plumber in Oakland, CA?"

We ran this question in Perplexity on September 21, 2026. The answer drew on 29 sources and recommended two plumbers. The first pick was a franchise location, which Perplexity called "the strongest verifiable all-around choice from the available information." Its reasons:

• 24/7 emergency service, stated on the location's own pages
• licensed and insured
• a flat-rate pricing policy with no overtime charges, stated in one sentence on the location page
• a California contractor license number, which sits in the site's footer next to the address

The second pick was a smaller Oakland-based shop with a 5.0 rating on directories. Perplexity added that review totals vary by directory.

The sources behind the answer were the two businesses' own sites, Yelp, Thumbtack, Nextdoor, LinkedIn, the California contractor license board, two local "best plumbers" roundups, and several small directory sites. Perplexity's suggested follow-up questions asked about exactly the facts a local business controls: which license the company holds, its exact hours and response times, and whether it charges extra for nights and weekends.

The takeaway for any local business: Perplexity recommended the company whose facts were easiest to verify, and it quoted those facts straight from the company's own pages. For the local version of this playbook, see Perplexity local optimization.

How to get cited by Perplexity

These steps are in the order they usually pay off. The Perplexity optimization checklist turns the page-level ones into a worksheet.

1. Make sure Perplexity can reach your pages

• Allow PerplexityBot in robots.txt. Perplexity recommends allowing it if you want your site to appear in its search results. Changes can take up to 24 hours to apply.
• Check your firewall and bot protection. If you use a CDN or WAF, allow Perplexity's published IP ranges for PerplexityBot and Perplexity-User. Perplexity's crawler docs include Cloudflare and AWS WAF steps. A "block AI bots" toggle can quietly undo every other step on this list.
• Put the answer in the HTML. Perplexity doesn't document whether its crawler runs JavaScript, so make sure the text that answers the question is in the page's HTML, not only rendered in the browser.

2. Answer the exact question in the first two sentences

Match the heading to the way people ask, for example "How much does a water heater replacement cost in Austin?", then answer it directly underneath, before any backstory. Perplexity quotes sentences, so write sentences that still make sense when lifted out of the page.

3. Replace adjectives with facts Perplexity can check

"Trusted, top-rated experts" gives Perplexity nothing to quote. These do:

Those are the same facts Perplexity pulled into the plumber answer and asked about in its follow-up questions.

4. Show who wrote the page and when it changed

Perplexity's source labels ask whether a site says who wrote each piece and whether it corrects its mistakes. You don't need a label to get cited, but the same signals help any reader trust a page: a named author or team, a visible "last updated" date, and a short note when something important changes. Dates count twice, because Perplexity stores a published date and a last-updated date for every source it returns.

5. Keep selling separate from answering

The labels also ask whether a site keeps news separate from advertising and opinion. For a business site, the equivalent is a guide that answers the question first and puts the offer in its own clearly marked section, instead of threading it through every paragraph.

6. Own the original version of your content

The Columbia Journalism Review's Tow Center found that Perplexity sometimes cites a syndicated or copied version of an article instead of the original (details below). If your content is republished elsewhere, publish on your own site first, have the copies point back with canonical links, and avoid pasting full articles onto other platforms.

7. Show up where Perplexity already looks

Ask Perplexity your five most important questions and open the Links tab. That list is Perplexity's reading list for your topic. For a local business, it is mostly directories, review sites, license boards, and local roundups. For software, it is docs, comparison listicles, review sites, and Reddit. Fix your listings there first, then work on getting into the roundups. A recommendation often comes from those sites, not from yours.

8. Write your title and meta description as the citation preview

Perplexity shows your title and meta description in the hover card and the Links tab. Make both say what the page answers and for whom. "Water heater replacement cost in Austin (2026 prices)" gives a reader a reason to click. "Home | Smith Plumbing" doesn't.

Perplexity citation tracking: how to see where you're cited

Analytics only sees the people who click. To see citations, you have to ask Perplexity the questions yourself, or read the traces it leaves on your server. Here are five ways, from free to paid.

1. A manual prompt log (free)

• Write 10 to 30 buyer questions, not brand searches: "best [service] in [city]", "how much does [service] cost", "[option A] vs [option B]".
• Run each question in a fresh thread, in the same mode, and note where you ran it from, since local answers change by city. Answers also vary between runs, so run each question two or three times.
• Log every run. An example row:

• Add two columns if you can: the competitors that were cited, and the third-party sites that appeared (Yelp, directories, roundups). Those tell you where to work next.
• Repeat every week or two and track share over time ("cited in 7 of 30 runs") instead of a single yes or no.

2. Perplexity's API (automated, under a dollar a month)

The API returns the same evidence in structured form:

• Agent API with web search. The response includes a search_results item that lists every source with id, url, title, snippet, date, and last_updated. The answer text marks citations like [1], which map to those ids, and the research presets (low, medium, high) use markers like [web:1]. Perplexity tells developers to treat search_results as the source of truth and never to trust URLs the model writes itself.
• Search API. It returns Perplexity's raw ranked results for a query, up to 20 per query and up to five queries per request. Use it for a narrower question: is my page in the candidate set at all?

On cost, Perplexity's pricing page lists a web search call at $0.0025 plus model tokens, and its own examples put a typical cited answer at about $0.006 to $0.007. Thirty questions a week is roughly 130 answers a month, so under a dollar. The Search API costs $5 per 1,000 requests.

Two caveats. First, the API is not the consumer app: the app picks its own models and modes, while the API runs the model or preset you choose. Use the API for trend lines and spot-check in the app. Second, if your scripts still call the older Sonar chat completions endpoint, move them. Perplexity supports Sonar only until September 27, 2026, and maps Sonar to the Agent API's fast preset and Sonar Pro to low.

3. Server logs (the signal most people miss)

Search your server or CDN logs for Perplexity's two user agents:

• Perplexity-User means a Perplexity user's question triggered a live fetch of that page. Perplexity says this agent visits pages to answer questions accurately and may link to the page it read in the answer. Group these hits by URL and you get a list of pages Perplexity is pulling into answers right now, without running a single prompt.
• PerplexityBot is index crawling. Regular visits mean your pages are eligible. If an important page never gets a visit, find out why.

Check the IP addresses against Perplexity's published ranges, since anyone can fake a user agent. Treat the counts as directional: Perplexity can cite a page from its index without fetching it again.

4. Referral traffic

When someone clicks a citation, the visit usually arrives as a referral from perplexity.ai, although the referrer isn't always passed, so treat the number as a floor. Filter sessions by that source and look at the landing pages: those are cited pages that also earn visits. How to measure Perplexity traffic covers tying those visits to leads and to the changes you shipped.

5. Tracking tools: what to ask before you pay

Dozens of tools now run prompts through Perplexity for you. Before paying for one, ask:

• Does it query the Perplexity app or the API, and in which mode or with which model?
• Can it set the location? Local answers change by city.
• How many times does it run each prompt? One run a week is a coin flip.
• Does it separate retrieved, cited, and recommended, or count every mention as a "citation"?
• Can you see the raw answer and the full source list, not just a score?

If a tool can't answer those, a spreadsheet and the API will tell you more.

Are Perplexity citations accurate?

Mostly, but not always, and the failure modes matter if Perplexity talks about your business.

The most-cited independent test is from the Tow Center for Digital Journalism, published in the Columbia Journalism Review on March 6, 2025. Researchers ran 1,600 queries across eight AI search tools, asking each one to identify the original source of a news excerpt. Together the tools answered more than 60% of queries incorrectly. Perplexity answered 37% incorrectly, compared with 94% for Grok 3. The study also found that:

• Perplexity sometimes cited syndicated or republished copies instead of the original. Perplexity Pro did this for three of ten articles from the Texas Tribune, one of Perplexity's own publishing partners.
• Premium versions gave more confidently wrong answers than the free ones, because they declined to answer less often.

The test ran in early 2025 on news articles, so read the numbers as a snapshot, not a current score. Perplexity's own help center asks users to double-check sources. A citation shows where a claim came from. It doesn't prove the page says exactly that.

When Perplexity gets your business wrong

1. Find the source. Open the Links tab and find the page the wrong fact came from. It's usually an old directory listing, an outdated page on your own site, or a copy of your content somewhere else.
2. Fix it at the source. Update the listing or the page. Perplexity reads the live web, so a corrected page can change later answers once it's crawled again.
3. Report the answer. Use the flag icon below the answer and include the thread URL and what's wrong. Perplexity lists misinformation and outdated information among the problems it wants reported.
4. Rate the source. Each source panel has a "Rate this source" option for feedback about that domain.

Should you block PerplexityBot?

For almost every business, no. Blocking PerplexityBot works against appearing in Perplexity's search results, and a robots.txt block doesn't stop Perplexity-User, which fetches a page when a user asks about it. Blocking is a publisher's decision about licensing and traffic. Perplexity now pays some publishers directly: its Comet Plus program shares subscription revenue with participating publishers based on human visits, search citations, and agent actions. A plumber, a law firm, or a software company gains nothing by being invisible there.

A 30-day plan

Check where you stand

Optimizer's free audit reads your site the way AI search does. It checks whether your robots.txt lets PerplexityBot and other AI crawlers in, whether your content only appears after JavaScript runs, and whether your pages state your hours, services, and locations clearly. To see whether Perplexity names your business for the questions your customers ask, run the free Perplexity check.

Run a free audit. No changes are made without your approval.

Next steps

• New to Perplexity? Start with Perplexity AI optimization, the overview of how it works and what to optimize.
• Page-by-page worksheet: the Perplexity optimization checklist.
• Local business: Perplexity local optimization.
• Clicks and leads: How to measure Perplexity traffic.
• Other engines: Perplexity vs Google and ChatGPT citations.
• Product: Perplexity AI optimization with Optimizer.

Sources

Perplexity, checked September 21, 2026:

• Help Center: How does Perplexity work?, Understanding source labels, What is an answer engine?, How can I report incorrect or inaccurate answers?
• API docs: Perplexity crawlers, Agent API web search, Streaming citation parsing, Search API, Pricing, Migrating from Sonar, Changelog
• Perplexity: Getting started (usage figure, citation UI), Introducing Comet Plus (August 25, 2025)

Independent research:

• Tow Center for Digital Journalism, AI Search Has a Citation Problem, Columbia Journalism Review, March 6, 2025

Our tests: two questions run in the Perplexity web app on September 21, 2026, signed in on the free plan: "who is the best emergency plumber in Oakland, CA?" and "perplexity citation tracking". The cited pages' titles, meta descriptions, and on-page facts were checked against the live pages the same day.