How AI reads this page
ChatGPT, Claude, Gemini and Perplexity don't look at colours, images or animations. They receive text, headings, links and structured data. Below is exactly what they read when they open “Comparison and “best X” pages: why AI cites them and how to do them well”.
The AI receives 1,860 words in 8 sections, 14 links and 3 blocks of structured data.
Text
The page content in Markdown, without menus or graphics. It is the format closest to how a language model reads text.
# Comparison and “best X” pages: why AI cites them and how to do them well > How much “best X” pages weigh in ChatGPT, Google AI Mode and Perplexity citations in 2026, why they are declining, and how to write a comparison AI engines use. URL: https://geosnap.ai/en/blog/pagine-confronto-migliori-x-perche-ai-le-cita Language: English Version IT: https://geosnap.ai/blog/pagine-confronto-migliori-x-perche-ai-le-cita Author: Rinald Sefa, CMO Geosnap · Category: GEO strategy Publisher: Geosnap (Maind Group S.r.l.), https://geosnap.ai ## In short “Best X” pages are the most cited format in AI answers to commercial questions, especially on ChatGPT. In 2026, though, they are losing ground: with GPT-5.6 their share nearly halved in favour of product pages, and lists a brand writes about itself get cited far less than third-party ones. The better plan is to get into other people's comparisons, look after your product pages and, if you write a list of your own, make it complete, current and honest. - In the Wix study of more than one million citations, listicles account for 21.9% of the total and 40.9% on commercial-intent questions. - The weight varies by engine: strong on ChatGPT, low on Google AI Mode, which favours the brand's official pages. - On ChatGPT, listicle citations fell 30% between December 2025 and January 2026 and, with GPT-5.6, their share went from 15.47% to 8.19%. - In professional services, 80.9% of cited lists are third-party: a ranking where you put yourself first is not very convincing. - Lists that keep growing have an updated year, 10-20 options, stated criteria and a structure with pros, cons, pricing and a “best for” line. Written by the Geosnap AI agent, reviewed and approved by Rinald Sefa. Translated from the Italian original. [Read the original](https://geosnap.ai/blog/pagine-confronto-migliori-x-perche-ai-le-cita) Ask ChatGPT "what is the best reservation software for a restaurant" and, next to the answer, you will almost always find a page called "The 10 best reservation tools of 2026". For years these pages were the most popular shortcut into AI answers: many brands wrote one on their own site and put themselves at number one. In 2026 the picture has changed. Lists still matter, but less than before, not on every engine in the same way, and almost never when you write them about yourself. Here is what the data says and what to do with it. ## Why do AI engines cite "best X" pages so often? Because they match the shape of the question. Someone asking "which is the best" wants a comparison between alternatives with clear criteria, and a list page already contains everything: the options, pros and cons, pricing, a "best for" line. For an AI engine that has to build an answer in a few seconds, it is the easiest source to summarise. The numbers back this up. The [Wix Studio AI Search Lab study](https://searchengineland.com/ai-citations-favor-listicles-articles-product-pages-study-472364), published in March 2026 and based on 75,000 answers and more than one million citations from ChatGPT, Google AI Mode and Perplexity, puts listicles first among cited formats: 21.9% of all citations, ahead of articles (16.7%) and product pages (13.7%). On commercial-intent questions the share of listicles rises to 40.9%, nearly double any other format. On informational questions, articles win with 45.5%. The study's conclusion is the point to remember: what decides which format gets cited is mainly the intent of the question, more than the industry and more than the engine. A comparison page helps when the person is choosing; it does not help when they want to understand how something works. ## How much do they matter on ChatGPT, Google AI Mode and Perplexity? A lot on ChatGPT, little on Google AI Mode. [Peec's analysis](https://docs.peec.ai/research/content-types-cited-most-by-engine) from February 2026, based on more than one million citations, measures for each format how many pages get cited several times within the same answer. The result changes from one engine to the next: | AI engine | Strongest format | What the data says | | --- | --- | --- | | ChatGPT | "Best X" listicles | Just under 20% of cited pages; 52% of them appear more than twice in the same answer | | Google AI Mode | The brand's own category and product pages | No format dominates; the best one, category pages, stays just under 10% | | Perplexity | Listicles and articles, at low rates | Only about 6% of pages reach the most-cited bucket; according to Wix about 17% of citations come from discussions and forums | In practice: if your customers ask ChatGPT for recommendations, comparison pages are still an important route in. If they use Google AI Mode, your site's official pages weigh more, as we explain in our article on [how Google AI Mode picks its sources](https://geosnap.ai/en/blog/google-ai-mode-fonti-come-farsi-citare). Measuring per engine, rather than on average, keeps you from investing in the wrong format. ## Is the window closing? What the 2026 data says Yes: on ChatGPT, lists have been losing weight since the start of the year, and the drop accelerated over the summer. Seer Interactive tracked [more than two million citations](https://www.seerinteractive.com/insights/the-listicle-window-is-closing-in-ai-search-30-decline-mom) on the same prompts from November 2025 to February 2026, counting as a listicle any URL containing "best" or "top". Lists made up 16.6% of ChatGPT citations, but between December and January their citations fell 30%, against a 22.7% drop in total citations. On Google AI Overviews, over the same period and the same prompts, the listicle share was flat or slightly up. The space on ChatGPT went to Wikipedia (from 3.6% to 5.9%) and to Reddit, which nearly tripled its share. The second shift came in August 2026 with GPT-5.6, ChatGPT's new default model. According to [Peec's analysis of the rollout](https://docs.peec.ai/research/gpt-56-product-pages-vs-listicles), the share of listicles among retrieved pages went from 15.47% to 8.19%, while product pages became the leading category at 16.39%, with a 119% increase in daily retrievals. The reason is technical: on purchase questions ChatGPT fires secondary searches aimed at manufacturers' domains, and reads the product page directly instead of the list that compares it. Comparison pages lost share, but in absolute terms they are read even more than before, because the model retrieves more sources per answer. The same principle applies to shopping inside ChatGPT. [OpenAI's documentation](https://help.openai.com/en/articles/11128490-shopping-in-chatgpt) explains that the products shown are chosen from structured metadata supplied by merchants and third parties, from other third-party content and from the model's own knowledge; merchants are ranked by availability, price, quality and whether they make or directly sell the item. These are not ads: ChatGPT Ads is a separate channel with its own rules. ## Your own "best of" list is worth less than one written by others Much less. In professional services, the Wix study counted that 80.9% of cited listicles were third-party and only 19.1% were published by a brand about itself. The reason is simple: a list where the author puts itself first is not a comparison, it is a sales page, and models prefer neutral sources when the question asks for a judgement. This does not make the page on your site useless, but it changes the priorities. Other people's lists come first: review sites, trade publications, independent consultants' blogs, forum threads. The work to get into them is the one described in our article on [Reddit, forums and reviews](https://geosnap.ai/en/blog/reddit-forum-recensioni-perche-ai-li-cita): collect real reviews, get known by the people who write comparisons, answer objections in public. Then, if you write a comparison page of your own, write it the way a third party would. ## How to write a comparison page AI engines actually use Make it complete, current, with stated criteria and a structure that can be read at a glance. In Seer's analysis, 78 listicles kept growing while the others declined, and they shared four traits: 1. **Visible freshness.** Pages with "2020-2025" in the title were dropped from answers, pages with "2026" grew. One heavily cited software comparison lost 73% of its citations in a month because it still carried the old year. Update the year, the date and, above all, the content. 2. **Broad coverage.** Declining lists compared 5-8 options, growing ones 10-20. A short list adds no information beyond what the model already knows. 3. **Data and method.** Growing pages cite external sources (ratings from review platforms, public data) or explain how they chose and ordered the options. 4. **Structure.** All of them used numbered lists with, for each option, features, pros and cons, pricing and a "best for" line. Add two rules that the 2026 data makes mandatory. First: include competitors honestly, with their strengths, and do not always put yourself first; a list like "your furniture e-commerce against nine others" convinces the reader and the model only if it is credible. Second: look after your product pages, which ChatGPT now reads first. Price, features, availability and correct markup, as we explain in our guide to [structured data for GEO](https://geosnap.ai/en/blog/dati-strutturati-schema-geo-quali-servono), count for more than another ranking. To know whether it is working, look at the sources, not just the mentions. In Geosnap's monitoring you can see, for each commercial question your customers ask, which pages ChatGPT, Gemini and Google AI Mode cite: if the answers show third-party lists you are not in, you have found the page to work on. If they only show competitors' product pages, the problem is your product page, not the lack of a ranking. ## Sources - [AI citations favor listicles, articles, product pages: study](https://searchengineland.com/ai-citations-favor-listicles-articles-product-pages-study-472364) (Search Engine Land (studio Wix Studio AI Search Lab), 2026-03-24) - [The listicle window is closing in AI search: 30% decline MoM](https://www.seerinteractive.com/insights/the-listicle-window-is-closing-in-ai-search-30-decline-mom) (Seer Interactive, 2026-02-25) - [Content types cited most by engine: benchmarks from over 1 million AI citations](https://docs.peec.ai/research/content-types-cited-most-by-engine) (Peec, 2026-02) - [GPT-5.6: product pages vs listicles in ChatGPT retrieval](https://docs.peec.ai/research/gpt-56-product-pages-vs-listicles) (Peec, 2026-08) - [Shopping in ChatGPT](https://help.openai.com/en/articles/11128490-shopping-in-chatgpt) (OpenAI Help Center, 2026-09) ## Frequently asked questions ### Is it still worth publishing a “best X” page on my site? Yes, but not as the first move and not as a sales page in disguise. First work on getting into comparisons written by third parties, then write your own list with real competitors, stated criteria and regular updates. A ranking where you always come first gets cited far less. ### Why has ChatGPT been citing fewer lists since August 2026? With GPT-5.6, ChatGPT runs more secondary searches aimed at manufacturers' domains and reads product pages directly. According to Peec's analysis, the share of listicles among retrieved pages fell from 15.47% to 8.19%, while product pages became the leading category. ### Do comparison pages work on Google AI Mode too? Less than on ChatGPT. In Peec's analysis no format dominates on Google AI Mode, and the most cited pages are the brand's own category and product pages. If your customers use Google AI Mode, start from your site's official pages. ### What should a good comparison page contain? An updated year and date, at least ten options, explained selection criteria and external data where it exists. For each option: features, pros and cons, pricing and who it is best for, in a numbered list that is easy to read. ### How do I know whether lists matter in my industry? Ask the AI engines the questions your customers ask when choosing a supplier and look at the cited sources. If third-party lists show up, those are the pages to work on; if only product pages show up, your product page is the priority.
Structured data
Schema.org data (JSON-LD) embedded in the page code. It tells the AI explicitly what the page is about and who published it.
BreadcrumbList describes the path through the site
{
"@context": "https://schema.org",
"@type": "BreadcrumbList",
"itemListElement": [
{
"@type": "ListItem",
"position": 1,
"name": "Home",
"item": "https://geosnap.ai/en/"
},
{
"@type": "ListItem",
"position": 2,
"name": "Blog",
"item": "https://geosnap.ai/en/blog/"
},
{
"@type": "ListItem",
"position": 3,
"name": "GEO strategy",
"item": "https://geosnap.ai/en/blog/#strategia-geo"
},
{
"@type": "ListItem",
"position": 4,
"name": "Comparison and “best X” pages: why AI cites them and how to do them well",
"item": "https://geosnap.ai/en/blog/pagine-confronto-migliori-x-perche-ai-le-cita"
}
]
}
BlogPosting describes the article: author, dates and sources
{
"@context": "https://schema.org",
"@type": "BlogPosting",
"@id": "https://geosnap.ai/en/blog/pagine-confronto-migliori-x-perche-ai-le-cita#article",
"headline": "Comparison and “best X” pages: why AI cites them and how to do them well",
"description": "How much “best X” pages weigh in ChatGPT, Google AI Mode and Perplexity citations in 2026, why they are declining, and how to write a comparison AI engines use.",
"inLanguage": "en",
"url": "https://geosnap.ai/en/blog/pagine-confronto-migliori-x-perche-ai-le-cita",
"mainEntityOfPage": "https://geosnap.ai/en/blog/pagine-confronto-migliori-x-perche-ai-le-cita",
"image": "https://geosnap.ai/img/blog/p2026-10-14-pagine-confronto-ai-en.webp",
"articleSection": "GEO strategy",
"wordCount": 1300,
"author": {
"@type": "Person",
"name": "Rinald Sefa",
"jobTitle": "CMO",
"url": "https://geosnap.ai/en/about",
"worksFor": {
"@id": "https://geosnap.ai/#organization"
}
},
"publisher": {
"@id": "https://geosnap.ai/#organization"
},
"isAccessibleForFree": true,
"abstract": "“Best X” pages are the most cited format in AI answers to commercial questions, especially on ChatGPT. In 2026, though, they are losing ground: with GPT-5.6 their share nearly halved in favour of product pages, and lists a brand writes about itself get cited far less than third-party ones. The better plan is to get into other people's comparisons, look after your product pages and, if you write a list of your own, make it complete, current and honest.",
"datePublished": "2026-10-14",
"dateModified": "2026-10-14",
"citation": [
"https://searchengineland.com/ai-citations-favor-listicles-articles-product-pages-study-472364",
"https://www.seerinteractive.com/insights/the-listicle-window-is-closing-in-ai-search-30-decline-mom",
"https://docs.peec.ai/research/content-types-cited-most-by-engine",
"https://docs.peec.ai/research/gpt-56-product-pages-vs-listicles",
"https://help.openai.com/en/articles/11128490-shopping-in-chatgpt"
],
"translationOfWork": {
"@id": "https://geosnap.ai/blog/pagine-confronto-migliori-x-perche-ai-le-cita#article"
}
}
FAQPage describes the frequently asked questions
{
"@context": "https://schema.org",
"@type": "FAQPage",
"@id": "https://geosnap.ai/en/blog/pagine-confronto-migliori-x-perche-ai-le-cita#faq",
"mainEntity": [
{
"@type": "Question",
"name": "Is it still worth publishing a “best X” page on my site?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Yes, but not as the first move and not as a sales page in disguise. First work on getting into comparisons written by third parties, then write your own list with real competitors, stated criteria and regular updates. A ranking where you always come first gets cited far less."
}
},
{
"@type": "Question",
"name": "Why has ChatGPT been citing fewer lists since August 2026?",
"acceptedAnswer": {
"@type": "Answer",
"text": "With GPT-5.6, ChatGPT runs more secondary searches aimed at manufacturers' domains and reads product pages directly. According to Peec's analysis, the share of listicles among retrieved pages fell from 15.47% to 8.19%, while product pages became the leading category."
}
},
{
"@type": "Question",
"name": "Do comparison pages work on Google AI Mode too?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Less than on ChatGPT. In Peec's analysis no format dominates on Google AI Mode, and the most cited pages are the brand's own category and product pages. If your customers use Google AI Mode, start from your site's official pages."
}
},
{
"@type": "Question",
"name": "What should a good comparison page contain?",
"acceptedAnswer": {
"@type": "Answer",
"text": "An updated year and date, at least ten options, explained selection criteria and external data where it exists. For each option: features, pros and cons, pricing and who it is best for, in a numbered list that is easy to read."
}
},
{
"@type": "Question",
"name": "How do I know whether lists matter in my industry?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Ask the AI engines the questions your customers ask when choosing a supplier and look at the cited sources. If third-party lists show up, those are the pages to work on; if only product pages show up, your product page is the priority."
}
}
]
}
Metadata
The information in the page header, which crawlers and engines read before the content.