How can a B2B SaaS company get cited more often in ChatGPT and Perplexity?

A citation in ChatGPT or Perplexity passes through four stages:

• Parametric: what the model learned in training, frozen at a cutoff. Nothing you do this quarter changes it.
• Access: whether a crawler may fetch and read your page.
• Retrieval: whether the engine pulls your page into its context for a given query.
• Generation: whether the model cites what it retrieved.

You control access almost entirely. You influence retrieval only partly, and you don't control generation. So the work runs in this order:

1. Put the facts buyers ask about (pricing, plans, integrations, security) in the HTML before JavaScript runs.
2. Let the search crawlers in, bot by bot, in robots.txt.
3. Confirm your CDN or firewall doesn't turn those crawlers away.
4. State specific facts in plain sentences an engine can quote.
5. Get named on the third-party pages engines already cite for your category.

Steps 1 to 3 remove blocks. Steps 4 and 5 give the engine material to work with. None of them promises a citation. ChatGPT decides per query whether to search at all, and OpenAI says it rewrites the user's prompt into its own targeted queries. For the prompts SaaS buyers actually ask, see AI visibility for SaaS.

Many SaaS sites build pricing pages, docs and changelogs as client-rendered apps. Vercel's crawler study looked at ChatGPT and Claude crawlers. It found they fetch JavaScript files (11.50% and 23.84% of requests) but don't execute them, so they can't read client-side rendered content. Independent re-testing reported in June 2026 found the same for OAI-SearchBot and PerplexityBot. Googlebot is the exception, because it renders JavaScript.

This has a practical consequence. A pricing table that loads from an API after hydration may not exist for the crawler at all. The engine then answers from older training data or from someone else's page about you.

The next move

• Fetch your pricing, plans and top docs pages with JavaScript disabled. Compare the text with what a browser shows.
• List every price, plan name and limit that appears only after hydration.
• Server-render or prerender those pages. Tabs, accordions and "load more" blocks that fetch content on click have the same problem.

A common SaaS mistake is a blanket "block AI bots" rule, often added by security or legal. robots.txt rules are per bot, and OpenAI says each setting is independent of the others:

• GPTBot governs training use only.
• OAI-SearchBot governs ChatGPT search. OpenAI states that sites opted out of it will not be shown in ChatGPT search answers, though they can still appear as navigational links.
• PerplexityBot indexes pages for Perplexity search.
• Perplexity-User fetches pages for a user's request. Perplexity says it generally ignores robots.txt.

You can block training and still allow search:

User-agent: OAI-SearchBot
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: GPTBot
Disallow: /

Blocking GPTBot does not remove you from ChatGPT search. Allowing it does not get you cited either. Add Claude-SearchBot if Claude matters to your buyers. Platform detail for each engine lives in getting cited by ChatGPT and getting cited by Perplexity.

robots.txt is only half the gate. OpenAI's help centre sets out two conditions for eligibility. You must allow OAI-SearchBot, and your host or CDN must allow traffic from OpenAI's published searchbot IP addresses. Bot-management rules, JavaScript challenges and rate limits can return a 403 or a CAPTCHA to a crawler that robots.txt allows. Your analytics won't flag it.

Defaults can also change without anyone on your team editing anything. On September 15, 2026, Cloudflare set Training and Agent crawlers to blocked by default on ad-displaying pages. The change covered new domains and existing free-tier customers.

The next move

• Request a pricing page and a docs page as each AI crawler user agent. Record the status code.
• Check that /robots.txt itself is reachable to those crawlers.
• Ask security to allow verified search bots by user agent and verified-bot programmes, not by IP lists alone.

The AI Crawler Checker requests a page as each AI crawler and shows which ones the firewall turns away.

For "best tool for X" prompts, engines often cite review sites, comparison articles and community threads instead of vendor pages. Being named on those pages is associated with appearing in the answer. It is an association, not a promised effect, and brand size explains a lot of it. Keep your listings on the comparison and review pages that already show up for your category current and correct, including real pricing. More in third-party mentions and AI answers.

On your own site, state specific facts in single sentences. Say your Team plan costs $20 per seat per month (a hypothetical figure). Write exactly that on a page crawlers can fetch. It gives the engine a sentence it can quote. The 252,000-trial factorial found that explicit price information and a recent timestamp helped consistently. That study tested pages already in the model's context, so it shows effects on use, not on discovery.

Much GEO advice has no evidence behind it. Ahrefs tracked 1,885 pages that added schema and found no major uplift in citations on any platform, with AI Overviews citations falling 4.6%. C-SEO Bench found most content-rewriting methods largely ineffective and often harmful. Google says Search ignores llms.txt, and no other engine documents reading it. Treat these as non-levers. That doesn't mean you should rip out markup you already have.

Measure each engine separately. ChatGPT and Perplexity use different crawlers, indexes and citation formats, so one blended number hides which gate is broken. The free checker on this site shows whether AI engines name you for up to three buyer questions, as a one-time snapshot. To decide between monitoring and fixing, read tracking tools vs fixing tools. To set a re-check rhythm, read what changes between checks.