// AI NEWS · PLATFORM

Study of 1 Million AI Citations Finds Listicles, Articles, and Product Pages Win More Than Half

Wix Studio's AI Search Lab analyzed 75,000 AI answers across ChatGPT, Google AI Mode, and Perplexity — and found that the page type that gets cited depends on what the searcher is trying to do, not on your industry or the model.

2026-09-04 · by Moe Ameen

What happened

Wix Studio's AI Search Lab analyzed roughly 75,000 AI answers and more than 1 million citations across three major AI search surfaces — ChatGPT, Google's AI Mode, and Perplexity — to see which kinds of web pages get pulled into AI-generated answers. Search Engine Land reported the findings on March 24, 2026.

Three page formats accounted for just over half of every citation in the sample. Listicles led at 21.9%, articles followed at 16.7%, and product pages took 13.7% — 52% of all citations across those three shapes. But the headline the study itself pushes is not the ranking; it is that the single strongest predictor of which format gets cited was the searcher's intent, not the industry or the AI model. Split the citations by what the person was trying to do and the picture changes sharply: for informational questions, articles dominated at about 45.5% of citations (roughly 2.7 times any other format); for commercial, comparison-style queries, listicles took about 40.9% (nearly double the next format); and for transactional or navigational queries — someone close to buying or looking for a specific brand — product and category pages together drew around 40%.

Two secondary findings are worth pulling out. In professional-services queries, third-party listicles earned about 80.9% of the citations against roughly 19.1% for self-promotional "best of" lists a brand published about itself — evidence that the engines lean toward neutral, editorial comparisons over a company ranking its own product first. And the engines differed: Perplexity uniquely favored discussion forums like Reddit (around 17% of its citations), while ChatGPT leaned toward articles and informational content. These are one vendor's figures over a defined window, and AI engines change their citation behavior often, so treat the exact percentages as directional and confirm any single number against the source before repeating it. The durable takeaway holds up across separate studies: AI search cites by page type matched to intent, and product and listicle pages are far more central to that than a blog-only content plan assumes.

Why it matters for creators

  • Match the format to the intent or you do not get cited. A great how-to article will lose a "best X for Y" query to a listicle, and lose a "buy / pricing" query to a product page. The same topic needs different page shapes for different questions.
  • Product pages are a citation asset, not just a checkout step. If 13.7% of all AI citations — and about 40% of buyer-intent ones — go to product and category pages, a thin, unstructured product page is leaving high-intent visibility on the table.
  • Listicles win the comparison, but usually the neutral third-party one. Self-promotional "we're the best" lists earned a fraction of the citations that editorial comparisons did, so earning a mention on someone else's ranked list matters as much as publishing your own.
  • Intent beats industry and model. Because query intent — not your niche or whether it's ChatGPT vs Perplexity — predicts the cited format, the strategy generalizes: cover each topic in the shape the question wants, across surfaces.
  • A blog-only content plan is optimizing the smaller slice. Articles are one of three winning formats, not the whole game; the plans that show up in AI answers spread across listicle, article, and product/commercial pages.

How to act on this with Kompozy

The instruction hidden in this study is a production problem: to be cited across intents you have to publish the same idea in several shapes — an editorial article for the informational question, a ranked listicle for the comparison, tight commercial copy for the buyer-intent query — and keep doing it as engines shift. That is exactly the multiplication [Kompozy](/) is built for. Point it at one source and its copy engine (Claude and OpenAI, governed by your [Persona Brief](/glossary/persona-brief)) generates the cited formats in parallel: a long-form [Blog Article](/glossary/output-buckets) structured with the clean H2/H3 hierarchy the cited pages share, a listicle or "how to choose" framework for the comparison queries, [Quote Graphics](/glossary/output-buckets) and [Carousels](/glossary/hyperframes) that name specific tools and numbers, and captioned [Persona Shorts](/glossary/persona-shorts) — then [Autopilot](/glossary/autopilot) publishes the set across the eight social platforms plus blog and email on a standing schedule. The study says structured, intent-matched, frequently-updated pages get quoted; the reason most teams never build that spread is that producing it by hand breaks by week three, and that is the part Kompozy removes.

Be honest about the boundary, because it is the useful part of the pitch. Kompozy does not build or optimize your storefront's product pages — that is your CMS or Shopify's job, and product-page citations are earned there. What Kompozy owns is the other 38% and the surrounding footprint: the article and listicle content, the owned blog, and the consistent cross-platform presence that turns your brand into a named entity the engines already recognize before they ever fetch a page. It also can't force an engine to cite you — no tool can. What it does is make the volume of structured, on-brand, intent-matched content the data rewards actually sustainable. For the related research on who gets cited and why, see our coverage of the [Meltwater–LinkedIn study on individual vs company-page citations](/news/meltwater-linkedin-ai-citations-study), [LinkedIn's own AI-visibility guidance](/news/linkedin-ai-visibility-content-guidance), and why [ChatGPT indexing your small site still isn't the same as citing it](/news/chatgpt-search-indexes-small-sites).

Quick takeaways

  • Wix Studio's AI Search Lab analyzed ~75,000 AI answers and 1M+ citations across ChatGPT, Google AI Mode, and Perplexity; Search Engine Land reported it March 24, 2026.
  • Listicles (21.9%), articles (16.7%), and product pages (13.7%) accounted for 52% of all citations.
  • Query intent predicted the cited format better than industry or model: articles won informational queries (~45.5%), listicles won commercial ones (~40.9%), and product/category pages took ~40% of transactional/navigational queries.
  • Neutral third-party listicles earned ~80.9% of professional-services list citations versus ~19.1% for self-promotional lists; Perplexity uniquely favored forums like Reddit (~17%).
  • The practical move is to publish each topic in multiple intent-matched, well-structured formats — which is what a content engine like Kompozy produces at volume.

Frequently asked questions

Which page types get cited most in AI search?

According to Wix Studio's AI Search Lab study of over 1 million citations, three formats made up 52% of all AI citations: listicles (21.9%), articles (16.7%), and product pages (13.7%). But the study's core finding is that the cited format depends on the searcher's intent — articles win informational questions, listicles win comparisons, and product/category pages win transactional or buyer-intent queries.

Do product pages really get cited by AI answer engines?

Yes. In the Wix Studio study, product pages earned about 13.7% of all AI citations overall, and product plus category pages together took roughly 40% of transactional and navigational queries — the ones where someone is close to buying or looking for a specific brand. A thin, unstructured product page underperforms; a well-structured one is a genuine AI-visibility asset.

Should I publish my own "best of" listicle to get cited?

It helps, but the study found neutral third-party listicles earned about 80.9% of professional-services list citations versus roughly 19.1% for self-promotional lists a brand published about itself. The engines lean toward editorial comparisons, so earning a mention on an independent ranked list matters at least as much as publishing your own.

How do I get cited across different query types?

Cover each topic in the format the question wants: an article for the informational version, a listicle for the comparison, and clear commercial copy for the buyer-intent query — all with structured headings and specific, dated detail. Producing that spread by hand is the hard part; a content engine like Kompozy generates the article, listicle, carousel, and social versions from one source and publishes them across a blog plus eight social platforms.

Related news

← All AI news · Get started →