How to Get Cited by ChatGPT and Perplexity: What I Found Analyzing Their Actual Sources

I run AEO/GEO experiments to understand how AI engines choose which sources to cite and recommend. This post is part of that ongoing research log: I tested real queries across ChatGPT, Perplexity, and Gemini, then recorded which sources appeared, which did not, and what patterns repeated.

What I did#

I ran real AEO/GEO queries through ChatGPT (GPT-5.5 instant, with search), Perplexity, and Gemini (3.5 Flash), and recorded every source each engine cited. It's a small first sample, but the patterns were consistent and immediate. One thing to flag up front: the model and mode matter as much as the engine. A lighter "instant" or "flash" model — or any model answering without live search — surfaces far fewer sources than a search-grounded answer.

Research note#

This was a small-sample AEO test, not a definitive industry benchmark. I tested 6 query/engine/mode combinations across 3 AI engines in June 2026 and recorded visible citations only. The goal was to identify early citation patterns that small sites can act on, then retest them over time.

Who this matters for#

This matters most for small SaaS companies, consultants, agencies, and niche publishers trying to appear in AI-generated recommendations before they have traditional SEO authority.

The data: who actually got cited#

The first thing that jumps out: small, specialist sites made up 7 of the 17 visible cited domains — sitting right next to HubSpot, Conductor, Search Engine Land, Coursera, arXiv, and Google's own docs.

Pattern 1 — The cited page exactly matches the query#

Every cited URL was a dedicated page whose title and slug mirrored the query: /best-aeo-tools, /get-cited-chatgpt, /seo-vs-aeo-2026. Not a broad "marketing" page that happened to mention the topic — a page built to answer that one question.

Pattern 2 — Small sites can get cited next to the giants#

cite.sh, getairefs.com, aiclicks.io, digitalotters.com, lasso-up.com, width.ai, devexhub.com — none of these are household names, and they were cited right alongside HubSpot and Conductor. If you're a small or new site, this is the headline: you can get cited now, without years of domain authority, if your page is focused, current, and well-structured. This is why AEO is worth testing before you assume the big domains have already won.

Pattern 3 — Format beats prose#

The cited pages led with a direct answer, then used tables, bullets, and "fast picks." The engines lifted that structure almost verbatim. Walls of paragraph text don't get extracted; standalone, quotable sentences do.

Pattern 4 — Whether you see citations depends on the model and mode#

The cleanest test is the same engine in two modes. ChatGPT 5.5 instant answered "AEO vs SEO" with zero sources. ChatGPT 5.5 Pro (thinking) answered the same kind of question and cited its sources — Coursera, the original arXiv GEO paper, and Google's own docs. Gemini 3.5 Flash, like the instant model, surfaced none. Perplexity cited on every query.

So it isn't just which engine — it's the model tier and mode. A search-grounded, deeper-reasoning model shows sources; a fast "instant"/"flash" model answering from memory often doesn't. That's the key mechanic 👇

The distinction most people miss: getting cited vs. getting remembered#

There are two different games:

1. Citation (retrieval). The engine searches the live web, answers from what it finds, and links the sources. Perplexity does this on every query by design. ChatGPT does it only when it decides the question needs fresh or specific information.
2. Memory (training). When ChatGPT answers from its training data (no search), it mentions brands it already "knows" — no links. Getting there means being mentioned so widely across the web that the next model absorbs you.

If you're starting out, play the citation game first. It's page-level and achievable now. The memory game is a slow, brand-ubiquity play for later.

One nuance that trips people up: visibility (you see a citation) is not the same as influence (your content shaped the answer). A model generates from learned patterns, not a clean "this fact → this source" map — so a page can shape an answer without being shown: pulled in a search but not surfaced, or absorbed during training. That means "ChatGPT didn't cite me, so GEO is pointless" is wrong. The visible citation is the measurable tip; the real footprint is wider. Chase the visible citation first — it's the part you can actually measure — knowing your influence runs deeper than what's shown.

Pattern 5 — Deeper models may reach for primary sources (early signal)#

The thinking model didn't just cite more — it cited differently. It pulled primary, authoritative sources (the original research paper, official Google docs, Coursera), while the lighter modes and Perplexity leaned on small specialist blogs. This is one data point, so treat it as a hypothesis — but if it holds, there are two tiers of citability: focused specialist content gets you into the lighter engines today; original, primary, authoritative work is what the deeper models reach for. The good news for any creator: original research — like the data in this post — aims at both.

The playbook#

1. Build one focused page per query you want to win — title and URL matching the query.
2. Lead with a direct answer. Structure it (tables, bullets, steps). Write sentences that stand alone.
3. Add original data or comparisons; add schema; keep the date fresh.
4. Test on Perplexity (it always cites, so the signal is clean). Don't test on ChatGPT-without-search — it cites no one, so you'll get a false negative.
5. Repeat the same test over time. AEO is not a one-time checklist; the useful edge comes from seeing which citation patterns keep showing up.

FAQ#

Does ChatGPT cite sources?#

It shows citations when it retrieves information at answer time — a live web search or documents. When it answers purely from its trained knowledge, it usually can't point to specific sources, because a model generates from learned patterns rather than a clean fact-to-source map. Important caveat: no visible citation doesn't mean your content had zero influence — retrieved-but-unshown pages and training data can still shape the answer; you just don't see the provenance.

Why does Perplexity always cite?#

It's built as an answer engine — it retrieves web pages first, then writes the answer from them, so every answer is grounded in citable sources.

Do you need a big site to get cited?#

No. In this analysis, small specialist sites appeared next to major domains in AI citations. A focused, current, well-structured page on the exact query can matter more than raw domain authority.

Is AEO the same as GEO?#

Essentially yes — Generative Engine Optimization (GEO) and Answer Engine Optimization (AEO) are used interchangeably for getting your content into AI answers. The tactics are identical.