Short answer: ChatGPT and Perplexity recommend software they find named, again and again, on pages they trust: comparison articles, "best tools" lists, Reddit threads and review pages. To get recommended, let their search crawlers in, describe your product in one consistent sentence everywhere, earn a place on the pages they already cite, and publish honest comparison pages of your own.
This guide is for software companies, not publishers. The question is not "how do I get my blog post cited?" (we covered that in why ranking on Google does not earn ChatGPT citations). It is "when a buyer asks for the best tool in my category, why does my product not make the list?"
When a buyer asks "what is the best CRM for a 20-person agency?", both tools search the web, read several pages, and build a shortlist from the products those pages mention. Your own website matters less than you might think: the shortlist mostly comes from what other people have written about you. Rarely mentioned products are rarely recommended.
Marketing an AI product? See marketing for AI startups.
One 2026 study makes the point well. DerivateX ran buyer prompts for 40 B2B SaaS categories through ChatGPT with web search, ten times each. When ChatGPT recommended a tool, it cited that tool's own website only 11.6% of the time. Most citations went to independent blogs and vendor content about the category; Reddit made up 8.4% and review sites such as G2 and Capterra just 0.9%. One study is not a law, but it matches what we see in our own tests.
Check this first, because it is the only step you fully control. ChatGPT search uses a crawler called OAI-SearchBot; Perplexity uses PerplexityBot. If your robots.txt or firewall blocks them, your pages cannot appear in their search answers. Both companies say changes take up to about 24 hours to apply. Our free AI crawler checker reads your robots.txt and shows which AI bots it lets in. An llms.txt file is a small extra, not a fix; if you want one, our llms.txt generator builds it in a few minutes.
| Bot | Used for | Block it and… |
|---|---|---|
| OAI-SearchBot | Showing sites in ChatGPT search answers | Your pages will not be shown in ChatGPT search answers |
| GPTBot | Training OpenAI's models | A separate choice; it does not control search answers |
| ChatGPT-User | Visiting a page when a user asks | robots.txt may not apply, because a person triggered it |
| PerplexityBot | Showing and linking sites in Perplexity results | You drop out of Perplexity's results |
| Perplexity-User | Fetching a page a user's question needs | Generally ignores robots.txt, because a person triggered it |
The most common problem we find is not robots.txt. It is a security tool such as a bot-protection setting on the CDN that quietly blocks these crawlers, or pricing and feature pages that only load with JavaScript. Ask your developer to check both.
Work in this order. Each step builds on the one before, and the first three cost almost nothing.
Because the AI has to match your product to the buyer's question. If your homepage says "revenue platform", your G2 profile says "sales engagement" and a review blog calls you "a CRM add-on", the model has three weak signals instead of one strong one. Pick the category buyers actually type, name who it is for, and repeat it everywhere.
A quick test: paste your homepage's first paragraph into ChatGPT and ask "what category of software is this, and who is it for?" If the answer is not the one you want, buyers' prompts will not find you either.
Find the pages first, then earn your place on them. Run your category prompts, note every source link, and sort them by how often they appear. Usually a short list of 10 to 20 pages, a mix of comparison blogs, "best tools" lists, Reddit threads and a couple of review pages, does most of the work for a category.
We helped build GoodFirms, a B2B software and services review platform with more than 100,000 company profiles and 70,000 reviews. The vendors that did well on category pages were rarely the loudest. They had complete profiles, a clear category, recent reviews from real clients, and someone who replied when the editorial team asked a question. Most vendors left their profile half-finished. That is still true on most directories and list articles today, and it is an easy gap to close.
We run the prompt test, list the pages ChatGPT and Perplexity cite for your category, and show you which ones mention you. Tell us your product category →
Pages that answer the exact questions buyers ask an AI tool: "X vs Y", "alternatives to Y", "how much does X cost", "does X integrate with Z" and "is X good for small teams". Write them plainly, with tables, real prices and dates. AI tools quote specific, checkable facts; they skip marketing claims they cannot verify.
Our view on comparison pages: be fair to the competitor. Say where they are the better choice. A page that admits "pick them if you need on-premise hosting" is far more likely to be quoted, and trusted by buyers, than one that calls every rival worse. It also protects you if the competitor's customers read it.
Perplexity always shows its sources, leans more on fresh pages and community threads, and lets you see exactly which pages shaped an answer. ChatGPT searches only when it decides a question needs it, and often names brands without linking them. In practice, Perplexity is the easier place to diagnose problems; the fixes are the same for both.
The two tools also draw on different kinds of sources, which we covered in why AI engines cite different sources. Because Perplexity lists its sources, start your source-page list there, then confirm the same pages appear in ChatGPT's search answers.
Use a fixed set of prompts and run each one several times, because answers change from run to run. We use a 20-prompt sheet: 10 category prompts ("best invoice automation software for logistics companies"), 5 comparison prompts ("Acme vs Rival"), and 5 problem prompts ("how do I stop chasing late invoices?"). Run each three times in a fresh temporary chat. If you later want a tool to repeat the sheet every month, compare trackers in our best AEO tools guide.
| Column | What to record |
|---|---|
| Prompt and run number | The exact wording, and run 1, 2 or 3 |
| Products named | Every product, in the order given |
| Are we named? | Yes or no, and position in the list |
| How we are described | Category and any wrong facts |
| Sources cited | Every URL shown |
Your headline number is simple: the share of the 60 answers that name you. If you are named in 6 of 60 today, that is 10%. Re-run the same sheet each month after one change. A move from 10% to 25% over a quarter is real progress; a single good answer is not.
Crawler fixes apply within about a day. Changes to your own pages show up once they are re-crawled, often within weeks. Getting onto the third-party pages that drive shortlists takes longest: plan on one to two quarters of steady outreach, reviews and comparison content before your LLM visibility (your share of answers) moves in a way you can measure. This is the work our SEO, AEO and GEO service covers for software companies. To see where you stand first, ask for a free AI visibility snapshot.
Implementation partner rather than a software vendor? Buyers ask different questions about partners. See how ERP partners get recommended by ChatGPT and Perplexity.
Not inside the answer. ChatGPT has tested ads since February 2026, but OpenAI says ads are labelled, shown apart from the answer and do not influence it. Perplexity phased out its sponsored answers. Paid listings on review sites stay labelled as paid. The reliable route is earned mentions on pages the tools already trust.
Not directly. OpenAI says GPTBot is used for model training, while OAI-SearchBot decides whether your site can appear in ChatGPT search answers. You can block one and allow the other. If you want to be shown in ChatGPT search, make sure OAI-SearchBot is allowed in robots.txt and not blocked by your firewall.
They help, but less directly than many guides claim. In one 2026 study of 40 B2B SaaS categories, review sites made up under 1% of ChatGPT's citations. Reviews still matter because list articles and buyers use them, and a steady flow of recent reviews makes you easier to include.
Usually because the pages it reads describe you wrongly or in mixed ways: an old category, outdated pricing or a feature you removed. Find the pages it cites for your brand, ask their owners for corrections, and make sure your own site states the current facts clearly near the top of the page.
They overlap. Both need crawlable pages and trusted mentions from other sites. The difference is that AI tools build one shortlist from several pages, so being named on comparison articles and community threads matters more than your own ranking. Good SEO is the base; AI recommendation adds the off-site work on top.
Monthly is enough for most software companies. Use the same prompt set each time, run each prompt several times, and change one thing between tests. Weekly testing mostly measures random variation in the answers, not real change.
Want to know which pages ChatGPT and Perplexity read for your category, and whether you are on them? See our SEO, AEO and GEO services, or tell us your product category and we will run the first prompt test with you.