LLM SEO: How to Show Up in ChatGPT and AI Search

What is LLM SEO, and what else is it called?

LLM SEO means optimizing a website, and its presence on other sites, so that AI assistants and AI search engines find it, understand it and mention or cite it in their answers. The LLM meaning is large language model, so the LLM in SEO is a new reader next to the classic search crawler. The same work goes by other names: LLM optimization or LLMO, SEO for LLMs, AI search optimization, generative engine optimization (GEO) and answer engine optimization (AEO).

So the LLM SEO meaning comes down to two halves: make your own pages readable and clear, and be on the other pages the engines read, since an answer cannot name your product unless those pages name it. Can ChatGPT do SEO for you? It can draft text, but it cannot put your product on the sites it reads: this guide is about being found by ChatGPT, not writing with it.

GEO vs SEO and AEO vs SEO: one job or two?

Generative engine optimization (GEO) and answer engine optimization (AEO) are the two names you meet most. Google's guide to optimizing for its generative AI features spells both out and adds: “These are both terms you may see used to describe work specifically focused on improving visibility in AI search experiences.” Its own answer to the GEO vs SEO question follows: from Google Search's perspective, “optimizing for generative AI search is optimizing for the search experience, and thus still SEO.” ChatGPT, Perplexity and Claude each run a search crawler too, so read LLM vs SEO narrowly: SEO vs LLM optimization is a change of target, not of method.

How AI answers find pages: the search crawlers

OpenAI, Perplexity, Anthropic and Google each publish their crawlers and keep two jobs apart: fetching pages for search answers and collecting pages for model training. Blocking the second does not remove a site from the first. How an answer picks its sources is not on these pages, so what follows is what each says about access.

How to rank in ChatGPT search and get cited by ChatGPT

OpenAI's crawler page treats each setting as independent: “a webmaster can allow OAI-SearchBot in order to appear in search results while disallowing GPTBot to indicate that crawled content should not be used for training OpenAI’s generative AI foundation models.” On search it is blunt: “Sites that are opted out of OAI-SearchBot will not be shown in ChatGPT search answers, though can still appear as navigational links.” So how to do SEO for ChatGPT, or how to rank on SearchGPT as many still type it, starts where OpenAI says it does: “To help ensure your site appears in search results, we recommend allowing OAI-SearchBot in your site’s robots.txt file and allowing requests from our published IP ranges below.” After that, how to rank higher in ChatGPT and how to get cited by ChatGPT come down to the two halves of LLM SEO and the checklist below.

Perplexity SEO: how to rank in Perplexity

Perplexity's page describes its crawler as a search crawler: “PerplexityBot is designed to surface and link websites in search results on Perplexity.” And it adds: “It is not used to crawl content for AI foundation models.” Perplexity SEO is the same work with PerplexityBot in place of OAI-SearchBot, at robots.txt and at the firewall, where Perplexity publishes rules for Cloudflare and AWS.

Claude: Claude-SearchBot and ClaudeBot

Anthropic's help article on its crawlers splits them the same way. On ClaudeBot: “When a site restricts ClaudeBot access, it signals that the site's future materials should be excluded from our AI model training datasets.” On Claude-SearchBot: “Disabling Claude-SearchBot on your site prevents our system from indexing your content for search optimization, which may reduce your site's visibility and accuracy in user search results.”

AI Overview SEO: how to rank in AI Overviews

Google needs no separate crawler for AI Overviews and AI Mode. Its page on AI features and your website says “AI is built into Search and integral to how Search functions, which is why robots.txt directives for Googlebot is the control for site owners to manage access to how their sites are crawled for Search.” Of Google-Extended, the token many robots.txt files block, Google's crawler list says: “Google-Extended does not impact a site's inclusion in Google Search nor is it used as a ranking signal in Google Search.” So AI Overview SEO is Google SEO, plus the Search Console setting in step 2 below, plus presence on the kind of pages AI Mode cited in our sample.

Can AI crawlers read your directory listings?

A directory listing is a page about your product on a site with its own crawl history, and an answer engine cannot use it unless its crawler may read that site. On 2 November 2026 we read the robots.txt of the 177 startup, SaaS and AI directories Submitator submits to and checked 11 crawler names against each. Nearly every directory lets the AI search crawlers in: of the 163 files we read, 160 are open to OAI-SearchBot, 160 to PerplexityBot and 161 to Claude-SearchBot, and the eight directories with no robots.txt block nothing.

The blocks that exist fall mostly on training crawlers: 24 of the 163 files block at least one and 16 block all five, and each file that shuts out a search crawler shuts out all five training names too. 16 of those 24 carry a block marked “BEGIN Cloudflare Managed content”, whose rules for every crawler carry the line Content-Signal: search=yes,ai-train=no,use=reference and which, under the comment “Training crawlers”, lists the crawlers it refuses: search yes, training no.

The best known of the three that shut out a search crawler is Indie Hackers, whose robots.txt turns away OAI-SearchBot, PerplexityBot and Claude-SearchBot along with the training crawlers. Googlebot, which feeds AI Overviews and AI Mode, may still read two of the three, Indie Hackers included.

At the other end, some directories ask for AI readers. OpenHunts, a launch platform we submit to (DR 72 in our catalog), heads a section of its robots.txt with the comment “AI crawlers explicitly welcome” and allows OAI-SearchBot, GPTBot, ChatGPT-User, ClaudeBot and PerplexityBot by name. It serves an llms.txt as well, and every launch plan on its pricing page, the free one included, lists an “AI/GEO-optimized detail page”. Its Premium Launch costs $9.90 and needs no badge.

What 27 Google AI Mode answers cited

Crawler access says who may read a page; citations show what an answer used. On 17 October 2026 we pulled Google's AI Mode answers for 27 queries from our market, from submit startup to directories to dofollow directories, and recorded every source link and every directory named. It is one engine on one set of queries, and Google notes that “AI Mode and AI Overviews may use different models and techniques, so the set of responses and links they show will vary.”

Answers named 16 directories we submit to and never linked 14 of them, Product Hunt included; the links went to lists, articles and threads. The two catalog directories among the sources were cited for pages of their own writing, not for listings: AI Tools Capital in one answer for its blog list of AI directories and its submission service page, and PeerPush in three for a post on its blog. So an LLM SEO plan puts getting into those lists next to the listings themselves.

LLM txt for SEO: what an llms.txt file does

llms.txt is a markdown file at the root of a site that gives language models a short guide to it and links to clean versions of its key pages. The proposal, by Jeremy Howard, first published on 3 September 2024 and revised on 10 August 2026, calls itself “A proposal to standardise on using an /llms.txt file to provide information to help agents use a website.” Google Search does not use it: its AI optimization guide says that creating such files “will neither harm nor help your site's visibility or rankings in Google Search, as Google Search ignores them.”

Directories have taken it up anyway: 76 of the 177 we submit to served an llms.txt on 2 November 2026, 43% of them. Uneed's points to a second, full file of its product directory, and FastPedia names the AI crawlers one by one in its robots.txt to let them in. So for LLM txt SEO: if agents use your docs or your product has many pages, an llms.txt is an hour of work for the agents that look for one, and it does not replace a crawlable site.

How to rank in ChatGPT, Perplexity and AI Overviews: an AI search optimization checklist

How to do LLM SEO, in order: the steps that decide whether an engine can use you at all and whether the pages it reads say what you want said. That is how to show up in AI search, and LLM SEO optimization for a startup is mostly these four steps done thoroughly: a workable LLM SEO strategy on their own.

1. Let the search crawlers in, at robots.txt and at the firewall

Allow OAI-SearchBot, PerplexityBot, Claude-SearchBot and Googlebot. Decide on the training crawlers separately: blocking them leaves search alone, though Google-Extended also covers grounding in Gemini Apps. Then check the layer robots.txt does not show: Google lists “Ensuring that crawling is allowed in robots.txt, and by any CDN or hosting infrastructure” among its basics, and Perplexity warns “If you’re using a Web Application Firewall (WAF) to protect your site, you may need to explicitly allow Perplexity’s bots to ensure they can access your content.” If your CDN wrote a managed robots.txt for you, read what it says.

2. Get indexed, and check Google's AI setting

Google's rule for AI Overviews and AI Mode is the rule for search: “To be eligible to be shown as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet, fulfilling the Search technical requirements.” Confirm your key pages are indexed, then open Settings in Search Console: its Search generative AI control manages your site's inclusion in these features, and Google says of including it: “This is the default control for all properties.” A new page deep on someone else's site can take a while to get indexed; how listing pages get crawled and indexed covers that part.

3. Write the facts in plain text, and the same everywhere

Say on the page, in words, what the product is, who it is for, what it costs and what it replaces. Then use the same name, one-line description, category and price on every profile and listing, so any page an engine reads about you says the same thing. A page title or a preview that says something else is worth fixing first; our link preview checker shows what yours says.

4. Be on the pages answers name and cite

In our sample, answers named launch platforms and directories and cited the lists, articles and Reddit threads around them. So list your product on the directories and launch platforms of your category, ask the authors of the comparison lists in your field to consider you and answer buyers' questions in threads as yourself. Google's guide warns that “seeking inauthentic "mentions" across the web isn't as helpful as it might seem.” That line is about mentions that pass for somebody else's opinion. A directory listing makes no such claim: it is your product's own entry, published by the directory under its own rules and shown as a listing.

LLM SEO tools: what an LLM SEO tool measures

LLM SEO tools fall into three groups, and judging any LLM visibility tool for SEO starts with what it can see. AI visibility trackers run a set of prompts through several answer engines and log which brands and sources come back: LLMrefs describes its product as “Track brand visibility, rank, citations. Benchmark competitors & optimize for keywords across ChatGPT, AI Overviews, Perplexity, Claude, Gemini + more generative answer engines”, and Profound as “See what your customers ask AI, deploy agents to act on it, and measure the results in a collaborative workbench”.

SEO suites with an AI module add the same kind of tracking to their keyword and backlink data. Semrush pitches AI Visibility with “Track your AI mentions and find opportunities for your brand to get more visibility across ChatGPT, Google AI, and other AI platforms.” Ahrefs Brand Radar opens with “Map your full AI funnel across 6 AI tools.”

First-party data is free: Search Console counts AI Overviews and AI Mode in your overall search traffic and adds a Generative AI performance report, and your server logs show which crawlers from the table above visit. A tracker reports what answered on the prompts it ran, a sample like our 27 AI Mode answers: use an LLM SEO tool for trends across many prompts, and judge a single answer as a single answer.

LLM SEO services and agencies: what you are buying

Whether an LLM SEO agency or an LLM SEO company sells it as ChatGPT SEO services or as AI search optimization, the work breaks into five parts: an access audit (robots.txt, firewall, indexing and the Search Console setting); content that answers buyer questions in plain text; outreach to get into comparison lists and articles; listings on directories, launch platforms and review sites; monitoring with one of the trackers above. When you compare LLM SEO services, ask which of the five a quote covers.

The listings are the part a submission service like ours does, with the clearest unit price: our comparison of directory submission services lays out who does it and for how much, and our guide to automating link work covers which parts software should do. Agencies can resell submissions under their own brand through our white label tier and run client launches from claude.ai, ChatGPT, Claude Code or Cursor through our MCP server.

Directory listings for LLM SEO, done for you

Directory listings are the part of LLM SEO you can order instead of earn, and the census above shows where they stand: nearly every directory we submit to lets the AI search crawlers read it. You pick yours in our dashboard, each with its DR beside it, from the catalog of startup directories we submit to, or by category on the AI tool directories page and the SaaS directories page.

Your name, links and screenshots go out from one profile to every submission, and each directory gets a description of its own, written to its length rules. Where a directory wants a badge, it comes from one component on your site, set up once, and we keep every badge in it live, so a new directory never means a commit. No directory enters our catalog until we have run a submission through it from start to finish.

From our terms: most directories are delivered in 24 to 48 hours, every directory within 21 days at most. If a directory rejects your submission, you get one more directory to pick by default (All Directories aside, since it already holds the whole catalog). A refund can be requested within 30 days of purchase, on the conditions in our terms.

You pay once: $29 for 30 directories, $49 for 60 or $79 for 100. All Directories costs $149, one time, for the whole catalog and every directory added to it later. The three plans each include free indexing by IndexMachine ($25 value): when a listing page with your link goes live, we push it through IndexMachine ourselves, with nothing for you to do. Google still decides what it indexes, and each AI engine what it cites.