Answer engine optimization: How to get cited by ChatGPT and Perplexity
Perplexity cites Reddit at 46.7% and blogs at 9.9%. Here's what determines which blogs make the 9.9% — and how to be one of them.
By Alex Diaz
Perplexity draws 46.7% of its top-10 cited sources from Reddit. Blogs and opinion content account for 9.9%. YouTube gets 13.9%.
Read that again. The world’s fastest-growing answer engine cites Reddit almost five times more than blogs. If your content strategy is “write blog posts and rank on Google” — you’re optimizing for a distribution channel that’s shrinking while ignoring the one that’s growing.
Key takeaways:
- Perplexity cites Reddit at 46.7% and blogs at 9.9% — write like the best Reddit answers: experience-first, specific, answer-leading
- Lead every section with the conclusion — LLMs extract the first 2-3 sentences
- FAQ sections with question-format headings match queries mechanically — highest-leverage AEO tactic
- Tables get extracted more reliably than prose — every comparison gets one
- JSON-LD schema,
llms.txt, and unblocked AI crawlers are the infrastructure layer
This is the tactical follow-up to Make Something Agents Want. That post covered the strategic shift — why AI agents and answer engines are the next distribution layer. This post covers the mechanics: what gets cited, what gets ignored, and how to structure content so LLMs extract and recommend your work.
Why does Perplexity cite Reddit more than blogs?
Reddit dominates AI citations for three reasons:
-
First-person experience. LLMs weight experiential content — “I did this and here’s what happened” — over abstract advice. Reddit is full of it. Most blogs aren’t.
-
Specific claims. Reddit posts contain exact numbers, named products, and verifiable details. “I pay $234/year in income tax in Canton Ticino” is citable. “Consider a low-tax jurisdiction” is not.
-
Structured Q&A format. Reddit’s question-and-answer structure maps perfectly to how LLMs process information: query → answer → supporting context. Blog posts that bury the answer under 500 words of setup get skipped.
The lesson isn’t “post on Reddit instead of blogging.” It’s write your blog like the best Reddit answers: experience-first, specific, answer-leading, with verifiable claims.
The AEO framework
Answer Engine Optimization (AEO) is to AI search what SEO was to Google. Different mechanics, same goal: get your content in front of people at the moment they’re looking for answers.
1. Answer first, always
LLMs extract the first 2-3 sentences that directly answer a query. If your opening paragraph is a windup about “the evolving landscape of international tax planning” — the citation goes to someone who leads with the answer.
Bad:
International tax planning is a complex and multifaceted discipline that requires careful consideration of various jurisdictions, treaty networks, and compliance requirements.
Good:
Switzerland has no CFC rules. If you own an offshore company making $1.2M/year and you’re Swiss tax resident, Switzerland cannot touch the company’s undistributed profits.
The good version gets cited. The bad version gets skipped. Every time.
Every section of every page should lead with the conclusion, then support it. Not the other way around.
2. Intent-matched headings
Your H2 should match the exact query someone types into ChatGPT or Perplexity. Not your clever headline. Not your brand-voice take. The literal query.
| Bad heading | Good heading |
|---|---|
| ”Our Approach to Tax Residency" | "Best Tax Residency for Entrepreneurs" |
| "Understanding CRS" | "Does CRS Apply to Crypto?" |
| "The Dominican Republic Option" | "How Much Does Dominican Citizenship Cost?" |
| "Distribution Matters" | "How to Get Your First 100 SaaS Customers” |
LLMs match headings to queries. If your heading contains the query verbatim, you’re more likely to be cited. This is the H2 equivalent of keyword matching — except the “search engine” is an LLM reading your content at inference time.
3. Quotable blocks
LLMs pull clean, self-contained statements with specific numbers. These are the sentences that get extracted and presented to the user as the answer.
Write sentences worth quoting:
- “RevenueHunt is used by 20,000+ Shopify stores with a 4.9/5 rating.” — Citable.
- “We help businesses grow.” — Not citable. Not specific. Not extractable.
- “The DR Golden Visa costs $200K and provides permanent residency within 3-6 months.” — Citable.
- “The Dominican Republic offers attractive investment opportunities.” — Worthless to an LLM.
Every paragraph should contain at least one sentence that works as a standalone answer if extracted from context.
4. FAQ sections with question-format headings
This is the highest-leverage AEO tactic. FAQ sections with question-format headings (H3s) match queries directly:
### How much does Dominican citizenship cost?
~$3,000-3,500 all-in, including government fees, document
legalization, translations, medical exams, and immigration lawyer.When someone asks ChatGPT “how much does Dominican citizenship cost?” — the LLM finds this heading, reads the answer below it, and cites it. The match is almost mechanical.
Every post on this site has an FAQ section at the bottom. Not as an afterthought — as a deliberate AEO strategy targeting long-tail queries.
5. Tables over prose
LLMs extract tabular data more reliably than prose. A comparison table is infinitely more citable than three paragraphs describing the same comparison.
| This gets cited | This gets ignored |
|---|---|
| Comparison tables with specific numbers | Paragraphs describing differences |
| Step-by-step lists with clear labels | Dense prose explaining a process |
| Feature matrices with yes/no/numbers | Marketing copy about capabilities |
Tables are a core element of the content on this site. Every comparison gets one. Not because tables look nice (they do) — because LLMs parse them directly and extract structured facts.
6. Structured data (JSON-LD)
Schema markup isn’t just for Google. LLMs and AI agents read JSON-LD structured data directly. Article schema, FAQ schema, HowTo schema — these tell machines exactly what your content contains.
Every blog post on this site has:
- Article JSON-LD with headline, author, dates, keywords
- BreadcrumbList for navigation context
- FAQ section targeting question-format queries
The site also has:
robots.txtallowing all AI crawlersllms.txt— a markdown summary of the entire site for LLM inferencesitemap.xmland RSS feed for discovery- All implemented using the /ai-rank skill
7. Trust signals
LLMs weight authoritative sources higher. Trust signals that matter for AI citations:
- Author attribution — a named person with a bio and verifiable identity
- Specific data — exact numbers, dates, amounts rather than vague claims
- Primary sources — links to government sites, official programs, original research
- Experience markers — “I did this” statements with specific, verifiable details
- Recency — published dates, “updated” timestamps, current-year references
“According to a recent study” is not a trust signal. “According to OECD CRS 2.0 framework effective January 2026” is.
What gets you blocked
Just as there are things that help, there are patterns that get you systematically ignored by answer engines:
- Gated content. If the LLM can’t read the page, it can’t cite it. Paywalls, login walls, and heavy JavaScript rendering that blocks crawlers make you invisible.
- Marketing language. “Best-in-class,” “world-class,” “revolutionary” — these are noise words. LLMs trained on billions of pages learn to skip them.
- No specifics. If your page contains zero numbers, zero dates, and zero verifiable claims, it has nothing to extract.
- Blocking AI crawlers. If your
robots.txtblocks GPTBot, ClaudeBot, or PerplexityBot — you chose invisibility.
The Reddit + blog strategy
The smart play isn’t Reddit OR blog. It’s both:
- Write the definitive blog post with structured data, FAQ sections, tables, and specific claims
- Answer questions on Reddit in relevant subreddits, linking back to the full post for depth
- Use Reddit language in your blog content — the phrases people use when asking questions are the queries LLMs match
Perplexity cites Reddit because Reddit has the answers in the format LLMs prefer. If your blog post reads like the best Reddit answer — experiential, specific, structured — Perplexity will cite you too.
Measuring AEO success
You can’t check your “AEO ranking” the way you check Google rankings. But you can measure:
- Search your own topics in Perplexity and ChatGPT. Are you cited? Which competitors are?
- Check your
llms.txtextraction. Feed your page to Claude or ChatGPT and ask it to extract the key facts. If the extraction is empty, your page is invisible. - Monitor referral traffic from AI sources. Perplexity, ChatGPT (via Bing), and other answer engines send identifiable referral traffic.
- Track citation mentions. Set up alerts for your brand name and key content being cited in AI-generated answers.
FAQ
What is Answer Engine Optimization (AEO)?
AEO is the practice of optimizing content so AI answer engines (ChatGPT, Claude, Perplexity, Gemini) extract, cite, and recommend it. It’s different from SEO — instead of ranking on a search results page, you’re being quoted in an AI-generated answer.
Why does Perplexity cite Reddit more than blogs?
Reddit content is experiential (first-person accounts), specific (exact numbers and named products), and structured as Q&A (query → answer format). Most blog content is abstract, promotional, and buries the answer under paragraphs of setup. LLMs prefer the Reddit format.
How do I know if my content is being cited by AI?
Search your topics in Perplexity and ChatGPT. Check if your site appears in the citations. Feed your page to an LLM and ask it to extract key facts — if the extraction is empty, the page is invisible to answer engines. Monitor referral traffic from AI sources.
Does AEO replace SEO?
No — it adds to it. Google still sends traffic. But AI answer engines are growing fast (34% of US adults use ChatGPT). Optimizing for both means: structured content that ranks on Google AND gets cited by LLMs. The techniques overlap — answer-first content, structured data, FAQ sections work for both.
What’s the most important AEO technique?
Answer-first content. Lead every section with the conclusion. If someone asks the question your heading implies, the first 2-3 sentences should be the complete answer. Everything after that is supporting evidence. LLMs extract those opening sentences — make them count.
This post is the tactical companion to Make Something Agents Want. That post covers the strategic shift. This one covers the mechanics. For an automated approach: the AI Rank skill on GitHub audits and optimizes content against both the LLM and AGENT frameworks.