What page types on a SaaS website get cited in AI answers?
Which SaaS page types get cited in AI answers, from listicles to docs and product pages: what 2026 studies show per engine, and what to build or fix first.Load more

On a SaaS website, the page types cited most in AI answers are informational articles, listicles, product and feature pages, documentation and the homepage. Pricing pages and case studies are rarely cited. The mix differs by engine: Google's AI Overviews favour informational articles, while ChatGPT, Gemini and Perplexity lean more on listicles and product pages.
Key takeaways
- In a September 2026 study of seven B2B SaaS sites, informational articles took 63.8% of Google AI Overview and AI Mode citations.
- Across ChatGPT, Gemini and Perplexity in the same study, listicles led at 23.1% of citations, followed by informational articles (15.4%) and product or feature pages (15.0%).
- Pricing pages (0.4%) and case studies (0.1%) were almost never cited by those assistants, even though buyers rely on them late in a decision.
- Each engine cites a different mix: ChatGPT cited documentation far more than Gemini or Perplexity, and Perplexity leaned hardest on listicles.
- Most B2B SaaS citations go to sites a vendor does not own: the ten most-cited domains carry over 35% of citations, and none is a software vendor's own site.
- Measure which of your own page types are cited, per engine, from a frozen prompt set before deciding what to build.
Which page types get cited most?
Informational articles, listicles, product and feature pages, documentation and the homepage account for most citations to SaaS websites. The most specific B2B SaaS data published so far is Position Digital's September 2026 study: seven B2B SaaS clients, 7,258 citations across ChatGPT, Gemini and Perplexity, plus Google AI Overview and AI Mode data, from March to August 2026.
Seven sites from one agency's client base is a small sample, and each site's existing content shapes what can be cited at all. Read the figures as direction, then check them against your own data.
| Page type | Google AI Overviews and AI Mode | ChatGPT, Gemini and Perplexity |
|---|---|---|
| Informational article | 63.8% | 15.4% |
| Listicle | 8.9% | 23.1% |
| Product or feature page | 0.7% | 15.0% |
| Documentation and help | 1.5% | 8.3% |
| Homepage | 9.9% | 8.3% |
| Templates and tools | 9.1% | Not reported |
| Comparison article | 2.4% | 1.0% |
| Pricing page | 0.1% | 0.4% |
| Case study | 0.1% | 0.1% |
Why does each page type get cited?
Each page type gets cited when it matches the shape of a question buyers ask an assistant. The reasons below are our reading of how retrieval works, drawn from running prompt sets for B2B SaaS categories, not findings from the studies, and the right-hand column is what makes a page of that type usable once it is retrieved.
| Page type | Why it gets cited | What makes it citable |
|---|---|---|
| Informational article | Answers a what-is or how-does question | The answer in the opening passage of each section |
| Listicle (best X for Y) | Matches discovery prompts that ask for options | Stated criteria, a fair write-up per option, a review date |
| Product or feature page | Answers qualification questions: capabilities, integrations, fit | Specifics written as text, one feature per section |
| Documentation and help | Precise, current answers to how-it-works questions | Public and crawlable, not behind a login |
| Homepage | Resolves what the company is and who it serves | A one-clause category description, consistent with your listings |
| Templates and tools | Matches practical template and calculator queries | The template or method readable as text on the page |
| Comparison article | Matches X vs Y and alternatives prompts | Fair, sourced claims about every option |
| Pricing page | Rarely cited, but read for pricing-model questions | The pricing model stated in text, even if prices are on request |
| Case study | Rarely cited: outcomes are hard to extract and check | Problem, method and a sourced result in the first passage |
How does the mix differ by engine?
Each engine cites a different mix, which is why the AI Visibility Measurement Standard reports platforms separately and never as a blended score. In the Position Digital data, ChatGPT cited product and feature pages (18.9%) slightly more than listicles (17.7%), and cited documentation at 14.4%, far more than Gemini (3.4%) or Perplexity (3.3%). Perplexity leaned hardest on listicles, at 31.9%.
A larger cross-industry study points the same way. Wix Studio's AI Search Lab analysed 1,056,727 citations across 75,000 answers from ChatGPT, Google AI Mode and Perplexity (March 2026). Listicles took 21.9% of citations, articles 16.7% and product pages 13.7%, with comparison pages at 2.2%.
| Page type | ChatGPT | Gemini | Perplexity |
|---|---|---|---|
| Listicle | 17.7% | 24.5% | 31.9% |
| Informational article | 12.0% | 16.7% | 20.2% |
| Product or feature page | 18.9% | 13.3% | 12.8% |
| Documentation and help | 14.4% | 3.4% | 3.3% |
| Homepage | 7.6% | 9.8% | 3.2% |
| Comparison article | 0.7% | 1.2% | 1.6% |
Do your own pages matter if most citations are off-site?
Yes, but they are the smaller part of the problem. Goodie's analysis of 5.7 million citations captured from February to June 2025 found that the ten most-cited domains in B2B SaaS prompts carry over 35% of citations, led by Reddit and G2, and none of the ten is a software vendor's own site.
Your pages decide whether you are usable once retrieved and what an assistant says when it cites you. Review platforms, community threads and publisher round-ups decide much of whether you are in the answer at all, which is what the citation sources programme addresses.
Build the page types that answer your buyers' questions, not the page type that topped a study. A listicle written only to catch citations, with no fair criteria, costs you the trust the rest of the site depends on.
What should a SaaS team build or fix first?
- Check that the page types you already have can be retrieved: public docs, crawlable product pages, and no snippet controls or crawler blocks left on templates.
- Put a self-contained answer in the opening passage of your informational articles and product pages, the types cited on every engine in the data. That is the core of answer engine optimisation.
- Write your pricing model in text on the pricing page, even if prices are on request, so assistants describe it correctly.
- Publish a fair, sourced comparison or best-for page only where buyers ask that question, with the criteria stated before the table.
- Make the homepage say what you are in one clause, consistent with your G2, Capterra and LinkedIn descriptions.
- Measure which of your page types are cited, per engine, then build more of what your buyers' prompts reward.
How do you find which of your pages get cited?
Run your frozen prompt set on each engine, record every cited URL, and tag each one with its page type. Count citations per page type per engine, with the number of answers and the window stated. How to monitor AI search visibility sets out the prompt set and the number of runs.
The result is your own version of the tables above. It is worth more than any study, because it reflects your category, your competitors and the content you actually have.
If you do not know which of your page types assistants cite, or why a competitor's pages are cited instead, Surge45 measures it per engine against the AI Visibility Measurement Standard, then builds the briefs, templates and workflows your team uses to restructure the pages that matter, inside your own accounts.
Frequently asked questions
Do pricing pages get cited in AI answers?
Rarely. In Position Digital's study of seven B2B SaaS sites, pricing pages took 0.4% of assistant citations and 0.1% of AI Overview citations. They still matter: an assistant asked about your pricing model needs a clear text description to get it right.
Are listicles the best page type for AI citations?
They were the most-cited type across assistants in two 2026 studies, but they are not the best page for every query. Google's AI Overviews cited informational articles far more in the B2B SaaS data. Build a listicle only where buyers ask a best-for question.
Does ChatGPT cite SaaS documentation?
Yes, more than other engines in the available data: 14.4% of ChatGPT citations to the seven B2B SaaS sites studied went to documentation and help pages, against 3.4% on Gemini and 3.3% on Perplexity. Public, crawlable docs are worth keeping that way.
Do case studies get cited in AI answers?
Almost never in the published data: 0.1% of citations in the B2B SaaS study. That does not make them useless, because buyers still read them, but they are not where AI visibility comes from.
Should we plan against these percentages for our own site?
Use them as direction only. The B2B SaaS study covers seven sites, and each study classifies page types differently. Your own citation data, from a frozen prompt set run per engine, is the number to plan against.
Sources
- Position Digital, What B2B SaaS content works best in AI search?, September 2026; seven B2B SaaS sites, 7,258 assistant citations, March to August 2026
- Wix Studio AI Search Lab, The content types most cited by LLMs, March 2026; 1,056,727 citations across 75,000 answers
- Goodie, The Most Cited B2B SaaS Domains in AI Search, September 2025; 5.7 million citations, February to June 2025
About Nathan Mzumara
B2B SaaS growth specialist · SEO, digital marketing and SaaS
Nathan Mzumara is a B2B SaaS growth specialist with a 12+ year career in SEO, digital marketing and SaaS. He writes for Surge45, offering groundbreaking B2B insights for B2B SaaS businesses on GTM engineering, AI search, attribution and pipeline. He is also the founder of Search Marketing Club, where he runs a community of search marketing professionals.
Related Articles
Get the next playbook first
New teardowns, playbooks and research for B2B SaaS revenue leaders, when they publish.




