How do I get cited by Perplexity specifically

You get cited by Perplexity specifically when Perplexity can retrieve a live URL and that URL states a lift-able answer with facts other pages do not share. Perplexity’s Help Center describes the product as searching the internet in real time and attaching numbered citations to original sources. There is no public citation rate, no paid inclusion button for ordinary sites, and no Search Console clone that reports “you were cited N times this week.” Allow PerplexityBot, put the answer in HTML, then log the numbered source list on a frozen prompt panel.

This spoke sits under the Answer Engine Optimization playbook. Unique numbers have their own method in original research for AI citations. Community corroboration lives in Reddit for AI citations. Lane work is on /visibility.

The short answer

• Treat Perplexity as a retrieval-first answer engine with numbered footnotes, not as a Google rank clone and not as ChatGPT Search.
• Eligibility starts at crawl: PerplexityBot builds the search index; Perplexity-User fetches a page during a live question. Perplexity publishes both user-agents and IP JSON.
• Selection starts at writing: first-screen answer, one citeable object, dated unique numbers, entity facts that match About and schema.
• Search, Pages, and shopping are different surfaces. A Merchant Program feed is not a Search citation. A retired “Create page” button is not a webmaster console.
• Measure with a frozen prompt panel and screenshots of the numbered source list. Do not invent a Perplexity citation rate for the deck.

Move	Do this	Do not tell leadership
Crawl	Allow PerplexityBot; verify 200s from published IPs	“We unblocked AI, so we will be cited”
Fetch	Expect Perplexity-User on live questions; it generally ignores robots.txt	“robots.txt is the whole firewall”
Page	Answer in HTML, not after a client-only render	“We shipped a React shell, we are done”
Surfaces	Log Search citations separately from Pages and shopping	“We are in Perplexity” as one KPI
Scoreboard	Frozen prompts, source URLs, brand mentioned yes/no	A vendor “AEO score” with no query list

Do this afternoon, in order:

• robots.txt has an explicit User-agent: PerplexityBot / Allow: / (or a tighter path allow on public money URLs)
• WAF/CDN is not 403ing Perplexity’s published IP ranges
• Homepage, one money URL, and About return the offer sentence in view-source HTML
• You have 15–25 frozen prompts a buyer would actually type
• You have run those prompts in Perplexity Search and written down numbered sources

If the first two boxes fail, you do not have a Perplexity content problem yet. You have a fetch problem.

What does “cited by Perplexity” actually mean?

A Perplexity citation is a numbered source attached to an answer, pointing at a real URL a reader can open. Perplexity’s Help Center is explicit: each answer includes numbered citations linking to original sources so a person can verify or go further. That is the unit. A brand mention in the prose without a source chip is a different row. A Perplexity Page that restates your category is a different product.

I have been SEO certified since 2021. The certification does not mint Perplexity footnotes. Fetchable HTML and a sentence worth quoting do.

Signal	What you saw	How to log it	What it is not
Numbered citation	Your URL in the source list / footnote	Citation win	A Google rank
Brand in the answer, no URL	Name appears; source is a third party	Mention win, citation miss	“We own the answer”
Competitor URL, your idea	They restated your public copy	Content leak / missing unique number	Proof the engine “hates you”
Shopping card	Product tile, merchant, or Buy with Pro path	Shopping surface	A Search footnote
Page (Perplexity Pages)	A Perplexity-hosted document	Pages surface	Evidence Search cited you

Operator rule: if you cannot screenshot the number next to your hostname, you were not cited on that prompt. Write that down. Do not upgrade a mention into a citation in the weekly Slack.

Procedure for one prompt:

1. Paste the frozen query into Perplexity Search on the same account class you will use all quarter (Free vs Pro changes depth; pick one and freeze it).
2. Screenshot the full answer plus the source list.
3. Record every cited hostname, whether your brand string appears, and whether your money URL appears.
4. Click your footnote if you have one. Confirm it is the URL you meant, not a stale press release.
5. Save the date. Re-run on the same weekday cadence.

Failure mode: a marketing deck that says “Perplexity visibility is up” because someone saw the brand in one Pro Search thread on a laptop that was logged into a Focus mode nobody else uses.

How does Perplexity retrieve and attach numbered sources?

Perplexity’s public story is retrieval, then a summary, then citations. The Help Center walkthrough is: interpret the question, search the internet, compile an answer, attach numbered sources. Their getting-started post repeats the same shape: real-time search with sources and citations included. That is product copy, not a ranking patent. Use it as an operator model.

Pro Search is a user setting, not a webmaster lever. Perplexity’s help text says Pro Search can run a deeper pass and lets the user pick a model. The roster on that page moves. Do not treat a help-center model list as a ranking factor, and do not write a “we optimized for model X” ticket.

Layer	What Perplexity says in public docs	What you control	Hedge
Query interpretation	Understands the question, not just keywords	Title, H1, and lead that match the question people type	Internals unpublished
Live search	Searches the internet in real time	Crawlable, fresh, dated HTML	“Real time” is not a recrawl SLA
Source selection	Cites original sources so readers can verify	Unique facts, clear attribution, third-party corroboration	No published scoring weights
Numbered footnotes	Standard on answers	A URL worth opening	Citation count per answer is not a public constant
Pro Search	Deeper research; user-picked model	Freeze account class in the panel	Model choice is the user’s, not yours

Independent overlap work is the reason you do not copy a ChatGPT win into this plan. Ahrefs’ June 2025 Brand Radar cut found 86% of top mentioned sources were not shared across ChatGPT, Perplexity, and Google AI Overviews in that month’s sample. Winning one surface is not winning Perplexity.

Historically, recaps said Perplexity leaned on Bing-class APIs. As of August 2026, Perplexity documents its own crawler and index. Treat “Perplexity = Bing” as stale shorthand. Keep Bing healthy if you also care about Copilot and ChatGPT Search. Do not use Bing Webmaster Tools as a Perplexity console.

Checklist before you argue about “the algorithm”:

• You can open the Help Center “how it works” page and quote it without adding weights
• You are not using a 2024 “Perplexity is just Bing” slide
• Pro vs Free is frozen on the panel
• Focus mode is frozen (or logged as its own column)
• You have not averaged Perplexity, ChatGPT, and AI Overviews into one percentage

If the page cannot be fetched, none of the later writing advice matters.

How do I pass crawlability for PerplexityBot and Perplexity-User?

Perplexity’s crawler documentation splits two user-agents. Read both. Mixing them up is how a “block AI training” legal review deletes you from Search citations.

PerplexityBot is designed to surface and link websites in search results. Perplexity says it is not used to crawl content for AI foundation models. They recommend allowing it in robots.txt and permitting requests from published PerplexityBot IPs. Full user-agent string: Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; PerplexityBot/1.0; +https://perplexity.ai/perplexitybot).

Perplexity-User supports user actions. When someone asks a question, Perplexity may visit a page to answer and include a link. Perplexity says this agent is not used for web crawling or for training foundation models. Full user-agent string: Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; Perplexity-User/1.0; +https://perplexity.ai/perplexity-user). IPs: perplexity-user.json. Because the fetch is user-requested, Perplexity says this fetcher generally ignores robots.txt.

Robots changes: Perplexity says each robots setting works independently and it may take up to 24 hours for systems to reflect changes. That is a processing window for robots directives. It is not a promise that a rewritten paragraph will be cited tomorrow.

Agent	Job (Perplexity’s words)	Respects robots.txt?	If you block it
PerplexityBot	Surface and link sites in search results	Yes	You opt out of the index Perplexity tells webmasters to join
Perplexity-User	Visit a page during a user question; may include the link	Generally no (user-initiated)	A WAF 403 still kills the live fetch even if robots.txt is open

Minimum robots.txt if you want Search citations:

User-agent: PerplexityBot
Allow: /

User-agent: Perplexity-User
Allow: /

Allowing Perplexity-User in robots.txt is hygiene. It does not override a firewall drop, and Perplexity already says the user fetcher generally ignores robots. The WAF is the real kill switch.

Procedure:

1. Fetch https://yoursite.com/robots.txt and search for PerplexityBot. A global Disallow: / under * without a later allow is a fail.
2. Fetch perplexitybot.json and perplexity-user.json. Do not paste a blog’s IP list from last year.
3. Grep origin and CDN logs for both user-agent tokens on /, /pricing (or your money URL), and /about.
4. Confirm status 200 and HTML, not a challenge page, not an empty app shell.
5. Re-check 24 hours after a robots edit. Then re-run the prompt panel. Do not expect a citation the same afternoon.

Operator rule: user-agent strings are spoofable. Perplexity’s own WAF guidance is to match user-agent and IP. Copying only the UA into Cloudflare is how scrapers wear the costume and how you later panic-block the real bot.

Why do WAF rules and JavaScript-only pages kill citations first?

Because eligibility is a fetch. Perplexity’s crawler docs tell you to whitelist at Cloudflare and AWS WAF with combined user-agent plus IP conditions, using the JSON endpoints as the source of truth, and to refresh those ranges. Changes “may take some time to propagate.” Monitor logs. That is their instruction, not an agency invention.

A 2023 “block all AI bots” snippet is the most common self-own I still see on otherwise decent sites. It was written for training crawlers. It also catches search crawlers if you were sloppy with tokens.

Failure	What logs show	What Perplexity can do	Fix
robots.txt Disallow for PerplexityBot	Bot stops or never starts	Stay out of the search index they describe	Explicit Allow
Cloudflare challenge / bot fight	403, 429, or JS challenge HTML	Index a wall, or nothing	Allow rule: UA contains PerplexityBot or Perplexity-User and IP in published ranges
AWS WAF rate limit	403 from the Web ACL	Same	IP sets from the JSON + string match on both UAs, higher priority than the block
Client-only render	200 with empty <div id="root">	Nothing to quote	Server-render the answer block
noindex / nosnippet on the answer	200 but snippet-hostile	Other engines suffer too; Perplexity still has less text to lift	Remove accidental robots meta on public URLs
Auth wall	401/302 to login	The public sentence does not exist	Put the citeable facts on a public URL

Cloudflare procedure from Perplexity’s WAF notes:

1. Security → WAF → Custom rules.
2. Condition: User Agent contains PerplexityBot OR Perplexity-User.
3. And: IP source is in the current JSON ranges.
4. Action: Allow, with priority above the catch-all bot fight.
5. Automate a refresh of the IP sets. Perplexity says the addresses update regularly.

Then prove the HTML:

• curl the money URL as a plain GET. The offer sentence is in the bytes.
• View-source in a browser with JS disabled. Same sentence.
• JSON-LD matches the visible sentence (legal name, offer, city).
• No “enable JavaScript” as the only body text.

I have shipped hundreds of production sites. The ones that fail Perplexity first are not “thin content.” They are 403s and empty shells.

What does an answer-first page look like for Perplexity?

Same job as any answer engine, with a Perplexity-specific test: the first 60–80 words must stand alone if a numbered footnote dumps a reader onto the page. Perplexity’s product copy says it compiles a concise summary. If your lead is brand mythology, the summary will quote a competitor who answered the question.

Write the page as if the footnote is the only click you get.

Page element	Perplexity-ready	Citation-hostile
H1	The question people type	“Ultimate guide” with no answer
Lead	Decision or definition in 60–80 words, dated if it includes a number	Mission paragraph, then a hero video
Citeable object	Table, numbered procedure, or bounded checklist	Mood essay
Facts	Name, offer, geo, pricing bounds that match About / schema	Slogan that conflicts with /pricing
Date	Visible updated date on time-sensitive claims	A copyright year in the footer only
Next step	One handoff to a protectable URL	Newsletter wall before the answer

Rewrite skeleton that only fits a Perplexity-citation URL:

1. Question-shaped H1 — the prompt, not a vibe.
2. Lead answer — the sentence you would be proud to see next to footnote [3].
3. One table or procedure — units, dates, constraints.
4. Unique number or method — see the next section. If you cannot name one, you are competing on phrasing.
5. Entity block — legal name, who it is for, where, what you do not do.
6. Primary source links — so Perplexity can verify instead of substituting a Reddit guess for your pricing.

Checklist per money URL:

• First 80 words answer the primary query without the brand origin story
• One table survives copy-paste into a spreadsheet
• Every number has a date and a denominator
• /about and schema do not contradict the lead
• The CTA is after the answer, not in front of it
• Internal links go to the pillar and to the corroboration spokes only when they help

Do not open a fifty-post calendar. Fix the URLs a buyer prompt would actually need: who you are, what you sell, the comparison, the price bounds, the constraint. That is usually five pages, not fifty.

Why do unique numbers get cited when generic explainers do not?

Because a numbered footnote wants a fact it can attach to one URL. “Many teams struggle with AI search” is a sentence fifty blogs share. “In our 2026 panel of 40 frozen prompts, 12 cited this hostname” is a sentence only you can own — and you should only publish it if you actually ran the panel. Invented n is how you contaminate answers and how you earn a lawyer.

The method lives in original research for AI citations. GEO (Aggarwal et al., KDD 2024) found statistics, quotations, and source citations lifted share of the generated answer on their lab metric. That is visibility inside an answer, not traffic, and not a Perplexity-published weight. Do not paste “+40% Perplexity citations” into a Q3 plan.

Claim type	Perplexity can footnote this	Perplexity will footnote someone else
Dated original n with method	Your research URL	A roundup that stole the number without a link
Public spec with a primary link	The spec, maybe you as explainer	The vendor doc, if your page is a paraphrase
Pricing band you actually charge	Your pricing URL	A directory that invented a range
“Industry-leading”	Nobody serious	A review site with a number
Recycled definition	Whoever said it first, cleanly	You, if you added nothing

Procedure for one citeable number:

1. Pick a question Perplexity already answers with mush (“How long does X take?”, “What does Y cost?”, “How often should Z refresh?”).
2. Collect a real n: your panel, your ops log, your ship count. Verified receipts I will put on a page: 500+ automations built, 20,000+ hours on agentic systems, 35,000+ hours saved for clients, hundreds of production sites. I will not invent a Perplexity win rate for your brand.
3. Publish n, date, population, and method in HTML next to the lead.
4. Put the same number on a methods paragraph, not only in an infographic.
5. Ask one third party to repeat it with your URL attached. A number that exists only on your site is easier to skip than a number the live web agrees on.

Checklist:

• The number is not in an image only
• The date sits in the same paragraph
• You can defend the method on a sales call
• You did not round a competitor’s study into “our research”
• The prompt panel includes the question that number answers

Operator rule: if two vendors say the same sentence, Perplexity has no reason to pick you. Give it a reason that looks like a table.

How do Search, Pages, and shopping differ?

They are different products. Mixing them is how a shopping feed ticket gets filed as “Perplexity SEO.”

Search is the answer engine this playbook is about: a question, a summary, numbered sources. That is the Help Center object.

Pages is a Perplexity-hosted document feature. Perplexity’s Pages help article (last updated July 16, 2026) said they were enhancing the Pages experience, that “Create page” was temporarily retired, and that “Convert to Page” would return with enhanced capabilities. As of that note, you do not have a stable public “publish a Page to get Search citations” workflow. Do not build a Q4 plan on a retired button. A Page is not a replacement for your money URL.

Shopping is a product surface. Perplexity’s Shop like a Pro post launched a Merchant Program so large retailers can share product specs for live details. Perplexity says the program is free, distinct from sponsored-question ads, and that joining increases chances of being a “recommended product” because the products will be in their index. That is Perplexity’s hedge: “increased chances,” not a citation rate. Benefits they list: payment integrations for Buy with Pro, free API access for a merchant-scoped search experience, and a merchant dashboard. If you do not sell SKUs, skip this row.

Publishers’ Program is a partner track, not a citation API. Perplexity’s Publishers’ Program announcement says they have included citations from day one, and that advertising on related follow-up questions will share revenue with publishers whose content was referenced. It also mentions partner API access so a publisher can build an answer engine that cites only their own content. That is a media partnership. It is not a self-serve “pay to be footnote [1]” button for a 12-page services site.

Surface	Public object	Who it is for	Citation meaning	Hedge
Search	Numbered sources on an answer	Anyone with a fetchable URL	The KPI this post measures	No public rate; no ordinary webmaster console as of August 2026
Pages	Perplexity-hosted document	Users inside Perplexity	Not your site citation	Create page retired in the July 16, 2026 help note
Shopping / Merchant Program	Product index + recommended product + Buy with Pro	Large retailers with live product specs	Product recommendation, not a Search footnote	“Increased chances,” Perplexity’s words
Sponsored related questions	Paid follow-ups	Advertisers	Ad unit	Distinct from Merchant Program, per Perplexity
Publishers’ Program	Revenue share when ads run on answers that reference you	Media partners	Partner economics	Not a guarantee of Search selection

Decision list:

1. You sell expertise, services, or software → Search citations on owned URLs. Stop reading shopping blogs.
2. You sell products at scale and can share a live catalog → Merchant Program is a separate ticket from the blog rewrite.
3. You are a newsroom with partnership staff → Publishers’ Program is a business-development conversation, not an SEO plugin.
4. Someone on the team wants to “make a Perplexity Page instead of a site” → No. The July 2026 help note is your excuse to refuse.

• Search panel exists (prompts + screenshots)
• Shopping work has an owner only if you have SKUs and a feed
• Nobody is waiting on “Create page” to ship HTML
• Ad sponsorship is labeled as ads in the report, never as citations

Why does Reddit keep showing up in Perplexity source lists?

Because Perplexity is citation-dense and retrieval-first, and community threads are cheap, opinion-rich passages for “who should I use?” prompts. Third-party studies keep finding Reddit among Perplexity’s most-cited domains. Exact percentages move. Do not tattoo one number into the strategy doc.

Tinuiti’s Q1 2026 citation work, written up on Search Engine Land, reported Reddit at 24% of Perplexity citations in January 2026 in that commercial-intent panel — and 0.1% on Gemini in the same write-up. Same domain. Different surface. Different weight. Semrush-class 2025 cuts put Reddit even higher on some Perplexity slices. Recycled posts still quote 46.7% as if it were a law. It is not.

The operating method is Reddit for AI citations: show up as a human, disclose affiliation, answer the question someone asked, and link only when your page is the primary artifact. Spam and vote farms fail. Most cited threads are not this week’s press release.

What you hope	What Perplexity actually has	What you do
Corporate blog ranks as the only source	A thread that already settled “who is good”	Earn the thread; keep the owned URL honest
A new Reddit post this morning	Old Q&A with a specific answer	Answer living questions; do not dump links
Karma as a KPI	A citeable sentence	Measure cited Reddit URLs in the panel, not karma
Astroturf	Risk of removal and a worse brand sentence	Disclose; stop if you cannot be useful

Procedure:

1. Run the frozen Perplexity panel. Count how often reddit.com appears as a numbered source.
2. Open those threads. Are they about your category, your city, your product class?
3. If you are absent, you have a corroboration gap — not a robots.txt gap.
4. If you are present as a sneak-advertiser, fix the disclosure before you post again.
5. Point people at the owned URL only when that URL holds the table or the number.

Checklist:

• Reddit rows in the panel have thread URLs, not just “Reddit showed up”
• Affiliation is disclosed on any comment you make for the brand
• The owned page still answers the question if the thread dies
• You are not buying upvotes
• Legal has read the FTC endorsement rules before anyone “seeds” a subreddit

Community proof and editorial proof do different jobs. Perplexity will use both. You should too.

How do Focus modes change who can even be cited?

Focus is a user filter on the candidate pool. Perplexity’s getting-started guide tells people to narrow results with Focus — Academic for journals, and similar source-type filters. If your buyer runs Academic, a vendor blog is not the pool they asked for. If they run a web-wide or social-weighted view, a journal PDF is optional.

You do not control the user’s Focus. You control whether you have a URL that belongs in the pool they might choose, and whether you log Focus as a column so you stop comparing two different products.

Focus / source filter (directional)	Pages that can still win	Pages that usually lose	How to log
Web / default Search	Fetchable, answer-first owned URLs	JS shells, blocked bots	Default panel row
Academic	Papers, .edu, standards, methods pages with citations	Thin listicles	Separate column
Social / community-weighted	Real threads, honest profiles	Corporate Reddit dumps	Separate column
Video	Transcript-backed video URLs if they rank in that pool	A homepage with no media	Only if buyers use it
Specific domain set	Whatever they scoped	Everything outside the scope	Do not treat as sitewide visibility

Procedure:

1. Pick the Focus your buyers actually use. Ask sales. Do not guess from Twitter.
2. Freeze it for 90 days on the panel, or run two frozen variants and never average them.
3. If Academic is the buyer habit, fund a methods page with primary citations — not another “what is AEO” post.
4. If community Focus is the habit, fund Reddit plus the owned table those threads should point at.
5. Re-screenshot after a Perplexity UI change. Focus labels move. Your log should name what you clicked, not what the blog called it in 2024.

• Panel header includes Focus / source filter
• Account class (Free / Pro) is in the same header
• You did not report a miss on Academic as a sitewide Perplexity failure
• Someone on the team can reproduce the click path from the log

Operator rule: a founder running Pro Search with Academic Focus is not your customer’s default. Stop using the founder’s laptop as the scoreboard.

What usually fails first when teams chase Perplexity citations?

Three failures, in the order I see them: the bot never got HTML, the HTML had no answer, the live web disagreed with the HTML.

The first is technical. The second is writing. The third is corroboration. Teams skip to “we need more blog posts” because that ticket is familiar.

Failure	Cost	What it looks like in the panel	What you do instead
PerplexityBot blocked at WAF	Zero chance at the index Perplexity describes	Competitors cited; you absent on every prompt	IP + UA allow; prove 200s
Answer lives behind JS	Fetch succeeds, quote fails	Random directory cited for your pricing	Server-render the lead + table
Generic explainer	You are interchangeable	A bigger publisher’s definition wins	Unique n + constraint table
Entity conflict	Model hedges or cites a stale PR	Wrong city, old package name, dead SKU	Align About, schema, pricing, LinkedIn
No community corroboration	Recommendation prompts skip you	Reddit / forums name competitors	Earn threads; do not spam
Scoreboard is vibes	Budget freeze or panic publishing	One screenshot, no frozen set	15–25 prompts, weekly

Concrete failure mode: a studio ships a beautiful client-rendered homepage, copies a “Disallow AI” robots snippet from a privacy blog, then spends six weeks rewriting blog titles. Perplexity never saw the new titles. The WAF returned a challenge page. The numbered sources on “best [category] partner” stay a mix of directories and a Reddit thread from 2023.

Recovery procedure:

1. Unblock PerplexityBot and Perplexity-User at robots and WAF. Wait the ~24-hour robots window Perplexity documents. Confirm logs.
2. Put the offer sentence in HTML on five URLs: home, money, About, proof/FAQ, contact or location.
3. Add one unique number you can defend.
4. Run the frozen panel. If you are still absent, read the sources that did win. They have a table, a date, or a thread you do not.
5. Only then open a content ticket that is not “publish 40 posts.”

• Last incident had a named cause (robots, WAF, JS, mush, conflict), not “Perplexity hates us”
• Recovery changed HTML or fetch, not just the title tag
• Panel re-run is dated after the fetch fix, not the same morning

Bravery is not a robots.txt strategy. Neither is a new editorial calendar on a 403.

How do I measure Perplexity citations without inventing a rate?

You measure inclusion on a frozen prompt panel. Perplexity does not publish a citation rate, a recrawl SLA, or a public Search Console equivalent for ordinary sites as of August 2026. Publisher partners may get extra analytics through program tooling Perplexity mentioned alongside ScalePost in the Publishers’ Program post. If you are not in that program, you do not have that dashboard. Do not fake one.

Ahrefs Brand Radar and similar vendor tools can show mention share across assistants. Use them for shape. They do not replace your numbered-source screenshots. Ahrefs’ overlap study is useful because it tells you not to average engines. It is not your weekly KPI.

KPI	Source	Cadence	Kill criterion
Citation rate on frozen prompts	Manual log: cited / prompts	Weekly	0% for a quarter on money prompts → fix fetch or reinvent the URL
Mention without URL	Same log	Weekly	Rising mentions, zero URLs → brand is leaking through third parties
Source mix	Hostnames in the footnote list	Weekly	Reddit-only on recommendation prompts → corroboration gap
Fact accuracy	Wrong price, city, package	Weekly	Any hallucination on a money fact → entity ticket same day
Referrer sessions from perplexity.ai	Analytics	Monthly	Directional only; Pew-class work on other engines shows in-unit clicks can be rare — do not make Perplexity your new #1 referrer plan
Protectable conversions	CRM	Weekly	The number that pays rent

Panel spec:

1. Freeze 15–25 prompts. Include who-you-are, what-you-sell, comparison, constraint, and one trap prompt that should not cite you.
2. Freeze device, account class, and Focus.
3. Run weekly for 12 weeks. Same weekday.
4. Log date, prompt id, cited URLs, your URL yes/no, brand string yes/no, hallucinated fact yes/no.
5. Report citation rate as your_url_cited / prompts_run for that week, with the prompt count in the denominator every time.

Do not:

• Average Perplexity with ChatGPT and AI Overviews
• Convert three founder anecdotes into “we have 40% share”
• Treat a Merchant Program dashboard as Search citations
• Call a Page view a site citation

I will not invent a “typical Perplexity citation rate” for your category. If a vendor sells you one, ask for the prompt list, the date, and whether Focus was frozen. If they cannot produce those, it is a slide, not a measurement.

What does a 30-day Perplexity citation sprint look like?

A sprint that ships fetch + five quoteable URLs + a panel. Not a new blog CMS.

Days	Action	Done looks like
1–3	Robots + WAF + log proof	PerplexityBot and Perplexity-User 200 on home, money, About; IP sets from the live JSON
4–7	Frozen panel v0	15–25 prompts, Focus and account class named, first screenshots
8–16	Answer-first pass on five URLs	Lead + one table each; view-source contains the offer
17–22	One unique number + methods paragraph	Dated n you can defend; third-party ask in flight
23–30	Panel v1 vs v0; source mix; entity mismatch list	Keep / fix fetch / reinvent decision, not a 40-post outline

Weekly after day 30:

Cadence	Owner	Artifact
Weekly	SEO or founder	10–25 prompt Perplexity log with hostnames
Weekly	Eng	Bot 200s / 403s for both user-agents
Biweekly	Content	One URL made more quoteable, or an explicit defer
Monthly	Marketing lead	Search vs Pages vs shopping vs ads, never one bucket
Quarterly	Leadership	Whether Perplexity is even a buyer surface worth the panel

Skip if you only have a week:

• Unblock the two user-agents at robots and WAF
• Put a 60-word answer on the money URL
• Run 15 prompts once and save the sources
• Do not start a Merchant Program application unless you already have a product feed team
• Do not wait on Pages “Create page”
• Do not pitch a Publishers’ Program as the week’s SEO task

When this is not worth doing yet:

• You cannot name a single buyer who uses Perplexity. Run a 20-prompt check anyway; if the category answers never include competitors you care about, keep a thin baseline and spend the sprint on the engine sales actually hears.
• The site is still a login-only app with no public facts. Citations need a public URL.
• You are in the middle of a legal block-all-crawlers rewrite and nobody has split training bots from search bots. Split them first.
• Shopping is the only Perplexity conversation in Slack and you do not sell products. Redirect that energy to Search pages.

If none of the five URLs can survive a footnote click, you are not in a Perplexity problem. You are in a website problem. The lane page is /visibility.

FAQ

How do I get cited by Perplexity specifically?

Allow PerplexityBot in robots.txt and at the WAF using Perplexity’s published IP JSON, put a lift-able answer and a unique dated number in HTML, and earn corroboration so the live web does not contradict you. Then log numbered sources on a frozen Search prompt panel. Perplexity’s Help Center says answers include numbered citations to original sources; it does not sell a public citation rate.

How do I measure whether Perplexity citations are working?

Run the same 15–25 prompts weekly in the same account class and Focus, and count how often your URL appears as a numbered source. Track mentions without URLs as a separate column. As of August 2026 Perplexity does not publish an ordinary-site citation console, so a vendor “AEO score” is not a substitute for screenshots.

What usually fails first when teams try this?

Fetch fails first: PerplexityBot blocked in robots or the WAF, or the answer missing from the raw HTML. Next is interchangeable copy with no unique number. Third is entity conflict and empty community corroboration. More blog titles do not fix a 403.

How long does this take to show results?

Perplexity says robots.txt changes may take up to 24 hours to reflect. That is not a citation SLA. Plan on proving fetch in days, shipping quoteable HTML in two to four weeks, and reading a 12-week panel before you call the channel a win or a waste. Re-check after the wait window, not the next morning.

What should I skip if I only have a week?

Skip Merchant Program paperwork, Publishers’ Program pitches, and anything that waits on Perplexity Pages’ retired “Create page” button. Unblock the crawlers, write one money-URL lead, run 15 prompts, and screenshot sources. If you sell SKUs and already have a feed team, shopping can be a parallel ticket — not the week’s only ticket.

When is this not worth doing yet?

Skip a full sprint when no buyer uses Perplexity, when you have no public URLs with facts, or when legal still blocks search crawlers and training crawlers as one blob. Keep a thin measured baseline if the category matters at all. Spend the deep work on the engine that already shows up in sales calls.

CTA

If you want Perplexity footnotes, stop buying a citation rate that is not for sale — unblock the bot, write a sentence worth numbering, and log the sources.

Lane: /visibility · Book a visibility audit.