How to Get Cited in Perplexity in 2026

Perplexity is a live-search answer engine that always shows sources. A citation there is a B2B shortlist event: the buyer sees your URL next to the claim, can click it, and can forward the answer with receipts. ChatGPT can name you without a link. Perplexity puts the source on the page.

That is not a vanity metric. G2's 2026 AI Search Insight Report, a March 2026 survey of 1,076 B2B decision-makers, found that 51% of software buyers now start research with an AI chatbot more often than Google, and 85% think more highly of a vendor an AI chatbot cites. 69% chose a different vendor than planned after chatbot guidance. When the engine that always shows sources names you, the buying group has a URL to check.

The mechanism is three parts: real-time retrieval (not training-memory), PerplexityBot and Perplexity-User actually fetching your pages, and extractable passages the engine can quote. This post is the Perplexity playbook. It is a sequel to Friday's GA4 measurement post and the Reddit citations post. It does not redo either. It does not recap Digital PR versus link building.

Austin Heaton runs this as Answer Engine Optimization (AEO) for B2B: get cited on the prompts that build a shortlist, then measure whether those citations become perplexity.ai sessions and pipeline.

How does Perplexity choose sources (live search vs ChatGPT memory)?

Perplexity searches the live web for the query, then writes an answer with numbered citations. ChatGPT can answer from model memory and only search when it decides to. Ranking on Google helps Perplexity more than it helps ChatGPT, but it still does not reserve a citation slot. You need a crawlable, extractable page in Perplexity's own index.

Perplexity's own description is blunt: it searches the internet in real time and every answer comes with clickable citations. Ahrefs, in an August 2025 Brand Radar study of 15,000 long-tail prompts, found Perplexity is the outlier among assistants. 28.6% of its cited URLs also rank in Google's top 10 for the same prompt. ChatGPT, Gemini, and Copilot sit around 8%. Ahrefs also notes Perplexity does not draw on Google or Bing's index. It has its own search index, built by PerplexityBot.

That split is the whole operating model:

• Live retrieval. The query fans out, pages are fetched, and the answer is grounded in those pages. A stale or blocked URL never enters the pool.
• Visible sources. Numbered citations sit on the answer. A buying group can click them. A mention without a URL is a weaker event here than on ChatGPT.
• Own index plus on-demand fetch. PerplexityBot crawls for the index. Perplexity-User fetches a page when a user's question requires it. Official docs say that user-initiated fetcher generally ignores robots.txt.

The GEO paper (Aggarwal et al., KDD 2024; Princeton, Georgia Tech, Allen AI, IIT Delhi) is still the best controlled test of what those pages should contain. On GEO-bench, adding citations, quotations, and statistics lifted source visibility by up to 40% on Position-Adjusted Word Count. The authors also ran the methods on Perplexity.ai and reported visibility improvements up to 37%. Keyword stuffing did not help. That is the content layer. Access and extraction come first.

Do not collapse this into "rank higher on Google." Most Perplexity citations are still not page-one URLs for the original prompt. Fan-out, recency, and extractable passages decide the rest. For Google's two surfaces, see Google AI Mode vs AI Overviews. For ChatGPT's agent workflow, see Deep Research citations. This post stays on Perplexity.

What should you put in robots.txt for PerplexityBot?

Allow PerplexityBot. Official Perplexity crawler docs recommend that if you want to appear in search results, and they publish the exact user-agent string and IP ranges. Perplexity-User is a separate, user-triggered fetcher that generally ignores robots.txt. A robots rule is not a WAF allowlist. Changes can take up to 24 hours.

From Perplexity's crawler documentation, fetched for this post:

• PerplexityBot surfaces and links websites in Perplexity search results. Perplexity says it is not used to crawl content for AI foundation models. Full user-agent: Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; PerplexityBot/1.0; +https://perplexity.ai/perplexitybot). IP list: perplexitybot.json.
• Perplexity-User visits a page when a user asks a question that needs it, and may include a link in the response. Full user-agent: Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; Perplexity-User/1.0; +https://perplexity.ai/perplexity-user). IP list: perplexity-user.json. "Since a user requested the fetch, this fetcher generally ignores robots.txt rules."

The allow group Perplexity recommends is the one that gets you into the index:

User-agent: PerplexityBot
Allow: /

Do not treat a Disallow for Perplexity-User as a lock. Official docs say that agent generally ignores robots.txt because the fetch is on behalf of a user. If you need to block user-triggered fetches, that is a WAF and IP decision, not a robots.txt hope. If you want to be cited, allow both at the edge.

WAF is where most B2B sites silently fail. Perplexity's guidance: match User-Agent and source IP from the published JSON, set Allow, and refresh those ranges. Cloudflare and AWS examples are on the same docs page. A robots Allow does nothing if the edge challenges the bot, the origin returns 403, or the page is a login wall.

A free AI SEO audit will tell you whether robots.txt blocks PerplexityBot. It will not tell you whether the WAF is still dropping the request. That check is logs plus the official JSON.

If PerplexityBot is allowed in robots.txt but never appears in your logs, the block is usually the WAF. Book a 30-minute call and we will read the live file and the edge rules together.

How do you write a page Perplexity can extract?

Write one question per H2, put a 40–60 word answer under it, then prove the claim with a statistic, a quote, or a table. Perplexity extracts passages, not brand slogans. Self-contained blocks survive retrieval. Hedged marketing copy does not. The GEO paper's winning methods were citations, quotations, and statistics — not keyword stuffing.

Perplexity needs a passage it can lift without the rest of the page. That is the same extractability standard Austin uses on product pages built for AEO, applied to comparison, alternatives, pricing, implementation, and proof URLs.

Build each commercial section like this:

1. Question H2. Match the buyer prompt: "What is [product] for [ICP]?", "How does [you] compare to [incumbent]?", "What does [plan] include?"
2. Answer capsule. 40 to 60 words. Named product, named ICP, one proof point. No "we are the leading." A model can quote this paragraph alone.
3. Evidence in HTML. A number with a date and a source. A short quote from a customer or a third party. A table with criteria, you, and the alternative. The GEO methods that moved Perplexity visibility were exactly these: Cite Sources, Quotation Addition, Statistics Addition.
4. One job per heading. Do not bury pricing inside a brand story. Do not mix "who it is for" with "who it is not for" in the same block. Fan-out queries retrieve sub-passages.

Tables beat paragraphs for comparison prompts. A three-column table (criterion, you, incumbent) is a passage Perplexity can cite as a source for a shortlist. A 400-word "why we win" essay is not. Put numeric prices or ranges in crawlable HTML. If the only commercial facts sit in a PDF, a modal, or a JS widget, Perplexity-User may fetch the URL and still have nothing to quote.

Keep claims consistent with G2, LinkedIn, and the rest of the site. Perplexity will open more than one source. If the product page says mid-market AP automation and the G2 profile says enterprise finance suite, the engine hedges or quotes the third party. That corroboration problem is the next section, not a reason to write more blog posts.

How do freshness, G2, and Reddit work as corroboration layers?

Perplexity prefers current pages, then checks whether a third party says the same thing. Refresh the revenue URL with new extractable facts. Keep G2 and Reddit as receipts, not as the homepage. Superlines' SE Ranking figures show two-month-old pages earn more AI citations than stale ones.

Freshness first. Superlines, re-fetched for this post, reports that pages updated within two months earn 5.0 AI citations on average versus 3.9 for pages older than two years — about 28% more. That is SE Ranking's study, quoted on Superlines' 2026 statistics page. A real refresh changes the table, the plan limit, the screenshot, or the dated statistic. A new CMS timestamp on last year's packaging is a fake bump. For the decay clock across engines, see why AI citations disappear after 30 days. Do not treat that post's half-life numbers as a Perplexity-only law. Use the two-month freshness gap as the operating window here.

G2 second. G2's same 2026 survey found 45% of B2B software buyers say a review-site citation is the most confidence-inspiring signal in an AI answer. Review sites were the #2 source influencing shortlists (43%), behind AI chatbots (54%). Perplexity will retrieve a G2 profile, a category grid, or a comparison page when the prompt is commercial. The playbook for that layer lives in how G2 reviews become ChatGPT citations. The Perplexity-specific rule is simpler: the G2 sentence and the on-site capsule have to match, or Perplexity will cite G2 and skip you.

Reddit third. Perplexity uses community threads as corroboration on "which tool is actually good" prompts. That is not a posting hack, and it is not this post. Why Reddit AI citations matter for B2B in 2026 covers the flywheel. Here, Reddit is one more retrieval path that should repeat the same use case, ICP, and competitor names as your comparison page.

Superlines' own 30-day sample (34,234 AI responses, January–February 2026) is a measurement warning: Perplexity produced about 20 times more website links than brand-name mentions. You can be cited as a URL and never named. Log the cited URL, not only the brand string.

If Perplexity is citing a competitor's G2 page or a year-old roundup instead of your comparison URL, book a 30-minute call. We will mark the corroboration gap on the live prompts.

How do you test whether Perplexity is citing you?

Run a fixed set of buyer prompts in Perplexity, log the numbered sources, and re-run two weeks later. One screenshot is not a test. Record the exact URL, not "we were mentioned." Cited-with-link, mentioned, absent, and misrepresented are different statuses. Perplexity's live retrieval means a second run can change the list.

Use 8 to 12 briefs that match how B2B buyers actually start. G2 found 33% of initial software-research prompts are category-based and 31% are competitor-based. Only 6% start with budget. Your set should look like that:

• Best [category] for [ICP]
• Alternatives to [incumbent]
• [You] vs [top two]
• Pricing for a stated team size
• Implementation / time-to-value
• Who is [product] not for

Log every run:

Field	What to record
Date and mode	When you ran it, and whether Focus was All, Academic, or another mode
Exact prompt	The full brief, including ICP and constraints
Your status	Cited with link, mentioned, absent, or misrepresented
Cited URL	Your page, a competitor page, G2, Reddit, or a roundup
Slot	Which numbered source you were, if cited
Claim	Accurate, outdated, or invented
14-day re-run	Still cited, rotated, or gone

Do not rotate the prompt set every week or you will never see a pattern. Do not mix Perplexity results into a ChatGPT screenshot folder. The engines retrieve differently, and your log will show a different URL mix. That is expected.

Tie the log to money the way Austin reports client work: 5,130 ChatGPT referrals, Lumanu's 101 conversions and 566 ChatGPT clicks, iSpeedToLead's 7.79% citation share, and Rise's 575% AI search expansion. Citation share and referred clicks are different columns. Do not squash them.

How do you measure Perplexity traffic in GA4?

Do not rebuild the measurement stack here. GA4's native AI Assistant channel still does not name Perplexity in Google's live definition, so Perplexity sessions with a referrer usually sit in Referral until a custom channel group catches them. The dual setup — native channel plus a custom group above Referral — is how to measure ChatGPT and Perplexity traffic in GA4, published Friday.

The Perplexity-specific reminder is short. Create the custom AI Assistants channel with a source regex that includes perplexity, reorder it above Referral, and split Session source so perplexity.ai is its own row. Mobile clicks can still land in Direct. Keep four columns: Perplexity citations, perplexity.ai clicks, conversions, and Google AI surfaces. One blended "AI traffic" number will hide this engine.

What does Austin Heaton do in a Perplexity citation sprint?

He baselines the live prompts, unblocks PerplexityBot, ships extractable revenue pages, aligns G2 and Reddit to the same claims, then re-tests in two weeks. He does not start a 40-post blog calendar. The person who finds the missing source ships the page.

The sequence, used when a B2B brand is absent from Perplexity shortlists:

1. Reproduce the gap. Run the 8–12 prompt set. Log who is cited and which URL holds the slot.
2. Fix access. robots.txt Allow for PerplexityBot, WAF allowlist against the official IP JSON, HTML that does not require a login to read the commercial facts.
3. Fix the page the query needed. Usually a comparison, alternatives, pricing, or product page with a 40–60 word capsule, a table, and a dated statistic. Not a thought-leadership post.
4. Align corroboration. G2 category language and quoteable reviews; Reddit only where a real thread already exists. Same ICP sentence everywhere.
5. Refresh, do not duplicate. Edit the live URL. Honest last-updated date. Same slug.
6. Re-test at 14 days. Log whether your URL entered the numbered sources and whether it survived a second run. Then look at the custom GA4 channel.

That order is how citation share becomes a system. iSpeedToLead's indexing and revenue pages came first; the brand now holds a 7.79% AI citation share, first in its set. Rise's same bottom-funnel bias produced a 575% AI search expansion. Blog volume was not the lever. Extractable commercial pages plus a crawler that can reach them were.

If you want that sprint run on your category prompts, not a slide about "AI visibility," book a 30-minute call with Austin Heaton.

Perplexity citation services from Austin Heaton

Austin Heaton is an independent SEO and AEO consultant who helps B2B, SaaS, and FinTech companies get named in the Perplexity answers their buyers already run. The work lives on his SEO and AEO services page. Clients work with him directly, so the person who reads robots.txt also ships the comparison page.

• Access baseline. PerplexityBot and Perplexity-User in robots.txt, WAF, and logs, checked against Perplexity's published user-agents and IP JSON.
• Revenue-page extraction. Comparison, alternatives, pricing, and proof pages with question H2s, 40–60 word capsules, and tables.
• Corroboration. G2 alignment and community receipts that repeat the on-site sentence, without turning this into a Digital PR-versus-links program.
• Prompt log plus GA4. Numbered-source testing in Perplexity, then the custom channel from Friday's measurement post so clicks and conversions have a home.

Execution typically begins within about 7 days. More on Austin Heaton. A free AI SEO audit checks whether PerplexityBot can reach the homepage. It will not run your buyer prompts.

Read Next:

• How to Measure ChatGPT and Perplexity Traffic in GA4
• Why Reddit AI Citations Matter for B2B in 2026
• How G2 Reviews Become ChatGPT Citations
• How to Get Cited in ChatGPT Deep Research Reports
• Why Your AI Citations Disappear After 30 Days
• Google AI Mode vs AI Overviews
• How to Build Product Pages for AEO

Frequently Asked Questions

Does ranking on Google get you cited in Perplexity?

No. Ranking helps more for Perplexity than for ChatGPT, but it does not reserve a citation. Ahrefs' 15,000-prompt study found 28.6% of Perplexity's cited URLs also rank in Google's top 10 for the same prompt, versus about 8% for ChatGPT, Gemini, and Copilot. Perplexity still uses its own index (PerplexityBot). You need a crawlable, extractable page, not only a blue-link position.

How many sources does a typical Perplexity answer cite?

Perplexity always shows numbered, clickable sources. The count varies by query and mode, and Perplexity has not published an official average. Treat "always sourced" as the product difference versus ChatGPT, which often answers without a visible source list. Log the numbered URLs on your own prompt set rather than assuming a fixed source count.

Do you need llms.txt to get cited in Perplexity?

No. Perplexity's official crawler documentation does not mention llms.txt. It tells webmasters to allow PerplexityBot in robots.txt, permit the published IP ranges, and expect up to 24 hours for robots changes to reflect. An llms.txt file is optional extra. It is not a substitute for Allow: / on PerplexityBot or for a WAF that actually lets the bot through.

How fast do page updates show up in Perplexity?

Robots.txt changes may take up to 24 hours, per Perplexity's crawler docs. A user-triggered Perplexity-User fetch can hit a live URL on the next query, because that agent retrieves on demand. Indexed ranking is slower. Superlines' SE Ranking figures show pages updated within two months earn about 28% more AI citations than old pages. Refresh extractable facts on the live slug; do not wait on a new URL.

Perplexity vs ChatGPT — which should B2B teams prioritize first?

Prioritize ChatGPT for buyer volume and Perplexity for visible-source shortlists. G2's 2026 survey found ChatGPT dominates every segment they measured, while 51% of B2B software buyers now start in a chatbot more often than Google. Perplexity is the engine that always shows sources, runs live retrieval, and overlaps Google more (28.6% top-10 URL overlap in Ahrefs). Run both on one entity graph. Sequence the work: access, extractable revenue pages, then engine-specific logs. Do not pick a single chatbot and ignore the other.