Welt Analytics Back to the site

How do B2B SaaS companies get recommended by ChatGPT and Perplexity?

SaaS, E-commerce, Branded

Quick Answer: B2B SaaS companies get recommended by publishing one page per buyer question (comparison, pricing and integration pages first) and measuring ChatGPT and Perplexity answers every week. Competitor named, you absent: publish that comparison page. Named but uncited: publish your own page. Named and cited: keep measuring. This applies to any software company whose buyers shortlist vendors in an assistant.

Key takeaways

  • Open your robots.txt and allow OAI-SearchBot and PerplexityBot before you write a single new page.
  • Lock the competitor list before your first share-of-voice reading and leave it alone for the whole series.
  • Give every comparison page at least one fact about each product that a reader can check at its source.

What gets a B2B SaaS company named in an AI answer?

In their search features, both assistants surface and link the websites their crawlers can reach. OpenAI's crawler page states: "OAI-SearchBot is used to surface websites in search results in ChatGPT’s search features" [1]. PerplexityBot does the same for Perplexity's results [2].

Perplexity's help center says PerplexityBot "will not index the full or partial text content of any site that disallows it via robots.txt" [3]. An old Disallow line for either crawler can hide your best comparison page.

B2B buying decides which pages matter. A Commerce Engine explainer on B2B SaaS notes that "purchasing decisions often involve multiple stakeholders" [4]. Each stakeholder's question lands on a different page.

Nothing in this loop guarantees a recommendation. Publishing raises the chance of being cited; measuring shows whether it happened.

What do ChatGPT and Perplexity answer for your buyers today?

Today's answers show which buyer questions have no page of yours. The usual error is a test prompt with your brand in it: "A prompt that names your brand does not show what a buyer who asks openly sees." [6] Write each prompt the way a buyer types it, with the category, a use case and a constraint.

In ChatGPT, a globe icon next to the response shows that search was used, and OpenAI tells users to click the citation links to review the original sources [5].

  1. Write 10 to 20 buyer prompts across comparison, how-to, pricing and integration questions, with no brand names.
  2. Run each prompt in ChatGPT. If the answer shows no globe icon, log the prompt as uncited.
  3. For answers with citations, click each citation link and copy the URL.
  4. Record whether your company is named, its position in the list, and which competitor pages are cited.
  5. Run the same prompts in Perplexity and record the same fields.

Expected result: a sheet with one row per prompt and engine: named yes or no, your position (for example 2nd of 5), and the cited URLs, with the questions that have no page of yours standing out.

Which pages should a B2B SaaS company publish first?

Start with comparison, how-to, pricing and integration pages, one per buyer question. Match each gap to its page:

What the prompt run showsPage to publishHow to confirm
Not named in the answer at allThe page for that buyer question: comparison, how-to, pricing or integrationRe-run the prompt at the next weekly review and note whether your brand is named
A competitor is named and you are notA comparison page stating a sourced fact about both products, with no disparagementRead the page the assistant cited for the competitor; at a later run note whether your page is among the citations
Named, but the cited page is not yoursYour own page on that exact question, so there is a first-party sourceThe cited URLs for that prompt include your domain
Named in ChatGPT, absent in PerplexityNo new page yet; re-run the prompt in Perplexity before changing pagesThe next Perplexity reading for the same prompt names your brand, or still does not
Named with a position or tone you do not wantAfter reviewing that prompt's sentiment and sources, a page stating the correct fact plainlyOn following runs, note whether the answer states the corrected fact

Each page needs a direct answer in its opening sentence, a number or a limit, and a named integration or plan. The guide on what content a SaaS company should publish to get cited by AI assistants covers formats in more depth.

The trap is one generic company page, or its bulk cousin. The same guide warns: "The most common mistake is the template page: dozens of near-identical pages with only the industry name swapped." [7] Give each gap its own page, written for that one question.

For a brand mentioned while other sources are cited, Profound's docs advise: "Build content that answer engines can cite." [8]

How often should a SaaS team measure AI answers?

Once a week: measure the answers, publish, then review the gaps. The piece on how a small SaaS marketing team can handle AI search optimization covers who does what.

Welt Analytics checks ChatGPT, Gemini, Google AI Overviews and Google AI Mode daily, and Claude and Perplexity weekly, so Perplexity changes show with a lag.

One run of one engine is a weak verdict: OpenAI describes ChatGPT search as prioritizing the latest available information [5], so judge a page on the full prompt set over several readings.

Share of voice is your brand's mentions divided by all brand mentions in the answers, times 100.

  1. Fix the prompt set and the competitor list for the whole series.
  2. Count your brand's mentions and each competitor's mentions across the week's answers.
  3. Divide your mentions by the total of all brand mentions and multiply by 100.
  4. Log the figure next to the previous week's figure for the same engines.

Expected result: one percentage per week on a fixed prompt set, so a change after a page goes live shows against the week before.

A large swing with nothing changed usually means the prompt set is too small, so add questions before trusting the trend. Adding a rival midway lowers your share even though no answer changed, and a single trend line that blends daily and weekly engines jumps each time the weekly ones refresh [9]. To benchmark against named rivals, see how to measure share of voice for your brand in AI answers versus competitors.

Worked example

Say a SaaS team tracks 15 buyer prompts and four competitors. In week one its brand is named 12 times and the competitors 36 times between them: 12 / 48 = 25%. In week four, after a comparison page goes live, the brand is named 15 times against 35 competitor mentions: 15 / 50 = 30%. With the set fixed, the rise reflects the answers; check the cited URLs before crediting the page.

Where does Welt Analytics fit for B2B SaaS companies?

Welt Analytics reports a brand's visibility, share of voice, position, sentiment, sources and competitors in AI answers, on the engine cadence in the previous section. On the Growth and Pro plans it writes the articles and publishes them on a blog on your subdomain, and its site states: "Every article we publish belongs to you." [10] There is a 7-day free trial with no card required, and the team scans your site and sets the first prompts.

As of 2026-11-13, excluding VAT, Starter is 89 EUR a month for 50 prompts, Growth 229 EUR with 1 managed article a day, and Pro 499 EUR with 3 a day; Enterprise pricing is agreed individually.

Weighing trackers? See the best AI search visibility tool for a B2B SaaS company.

To see this week's answers for your own buyer prompts without the spreadsheet, start with Welt Analytics.

Questions

Can a B2B SaaS company pay to be recommended by ChatGPT or Perplexity?

No. A recommendation is earned, not bought, and no one can guarantee one. What a team controls is crawler access, one clear page per buyer question and weekly measurement of the answers.

Do we need a different page for ChatGPT and for Perplexity?

No. One page per buyer question serves both engines. Expect Perplexity to confirm a win in your reports later than ChatGPT, purely because of how often it is read.

Will allowing these crawlers let them train models on our content?

For Perplexity, its crawler docs say PerplexityBot is not used to crawl content for AI foundation models [2].

Can we start without buying a tool?

Yes. A spreadsheet works while the prompt set is small and one person runs it. A tracker becomes worth it when you want daily history, several brands, or more engines without someone retyping prompts.

Sources

  1. OpenAI, "Overview of OpenAI Crawlers"
  2. Perplexity, "Perplexity Crawlers"
  3. Perplexity Help Center, "How does Perplexity follow robots.txt?"
  4. Commerce Engine, "What is B2B SaaS"
  5. OpenAI Academy, "Search and deep research"
  6. "What sources does ChatGPT use when it recommends products in my category"
  7. "What content should a SaaS company publish to get cited by AI assistants"
  8. Profound docs, "Interpret answer engine insights"
  9. "How do I measure share of voice for our brand in AI answers versus competitors"
  10. Welt Analytics