Self-Promotional Listicles in AI Overviews: Do Best Of Lists Still Get Cited?

A self-promotional listicle is a "best X" article in which the company that publishes it ranks itself or its own product in the top spot, and Google AI Overviews, the AI-generated summaries Google shows above or between search results, cite these lists less often than they did a few months ago. In a study of 100 "best B2B software" queries, Lily Ray found AI Overviews cited 38 percent fewer of these lists in September 2026 than in June 2026, while listicles as a format still earn a large share of AI citations.

This guide, part of the generative engine optimization hub, works as a diagnostic for self-promotional listicles and the other "best X" pages you already have. Generative engine optimization (GEO) is the practice of making pages easy for AI systems to find and cite, and a citation is a link to a page shown as a source inside an AI answer. The short version: keep comparisons that show real evidence, and rewrite, merge or retire lists that exist only to rank yourself first.

What did Lily Ray's listicle study measure, and where does it stop?

Lily Ray's study measured how often Google AI Overviews cited self-promotional listicles across 100 "best B2B software" queries and found 38 percent fewer citations in September 2026 than in June 2026. It is one consultant's measurement of one vertical on one AI surface, so it shows a direction, not a rate for your pages.

Ray, an SEO consultant, published the numbers in a Substack essay titled Prediction: the next massive Google update is around the corner on October 4, 2026. When AI Overviews did cite a self-ranking list, they left the publishing brand out of the recommendation 83 percent of the time, up from 69 percent in June. She writes that "in 70% of cases, the listicle is now helping competitors get recommended while the brand that published it is left out." The page still feeds the answer, but the answer recommends someone else.

The method has clear limits. PPC Land, which covered the update on October 5, 2026, put it plainly: "These are a consultant's own measurements, on a small sample from one vertical and one AI surface. The essay does not set out how recommendations were classified or how the June and September query sets were matched." Ray does not publish the query list or the classification rules, so the numbers cannot be reproduced from the outside. Ray also predicts that a large Google update will hit this tactic, writing that "it's a matter of when, not if", which is an opinion, not a dated announcement.

The October numbers build on her earlier essay, Is Google Finally Cracking Down on Self-Promotional Listicles? from February 3, 2026. There she looked at seven company blogs with dozens to hundreds of self-ranking lists each and reported drops of 29 to 49 percent in Sistrix visibility, a search visibility score from the SEO tool Sistrix, between mid January and early February 2026. She called self-promotional listicles "only one tactic among many" used by the affected sites, and noted that some sites with heavy blog declines "were not over-utilizing self-promotional listicles." Those drops can have several causes at once.

Do other citation studies show listicles still get cited?

Yes. Several 2026 studies show listicles remain one of the most cited page types in AI answers, especially for "best" and "top" prompts. Unlike Lily Ray's study of self-ranking lists, most of these studies count all listicles together and do not separate lists that rank the publisher first from independent roundups.

• Prosperity Media analyzed 122,787 citations across 467 prompts and found listicles in 421 of those prompts, or 90 percent. On best and top prompts, listicles took 49.3 percent of citations against 4.1 percent for brand sites, according to its GEO citation analysis from September 2026. The study notes it does not isolate brand-owned lists.
• Grow and Convert studied 15,000 AI answers for one B2B software client and found 85 percent of citations went to third party pages in its 2026 citation research. Yet AI Overviews mentioned the brand in 90 percent of answers that cited the brand's own page, against 25 percent when a third party page was cited.
• Scrunch, an AI visibility vendor, found that a cited self-promoting listicle raised the author brand's recommendation rate from about 4 to about 7 percent across seven AI surfaces, with competitors recommended instead in 24.3 percent of responses, as reported by PPC Land in September 2026. PPC Land notes the causal figures rest on about 6 percent of the data.
• Seer Interactive saw ChatGPT listicle citations fall 30 percent from December 2025 to January 2026, which the author calls a one-month comparison, in The listicle window is closing in AI search.

Taken together, these studies point the same way: lists still get cited, but the publisher of a self-ranking list gains little from it.

Self-promotional, editorial or third party list: which one gets cited?

A self-promotional list ranks the publisher first, an editorial comparison weighs named options on stated criteria and may still be published by a vendor, and a third party list is written by an independent site. Current studies suggest independent and evidence-led lists hold up in AI answers, while self-ranking lists increasingly help competitors instead of their publisher.

An editorial comparison is a page that compares named products on criteria the reader can check, shows how each one was tested or judged, and states where the publisher's own product is weaker. Here is what the 2026 evidence says about each type.

Criterion	Self-promotional list	Editorial comparison	Third party list
Who wrote it	The vendor, ranking itself first	The vendor or a publisher, ranking on stated criteria	An independent site
Evidence Google's review guidance asks for	Usually missing	First-hand tests, drawbacks, use cases	Varies by publisher
AI Overview citations in 2026	Down 38 percent from June to September 2026 in Ray's 100-query sample	Not measured separately in the studies reviewed	Most citations went to third party pages in Grow and Convert's 2026 data
Who gets recommended	Publisher left out 83 percent of the time in Ray's September 2026 sample	Depends on how well the page matches the prompt	Brand mentioned in 25 percent of AI Overview answers citing such pages (Grow and Convert, 2026)
Main risk	Feeds competitors, possible future update	Costs real testing time	You do not control the content

The practical reading: you cannot rely on your own list to put you first in an AI answer, and you cannot rely only on others' lists to mention you. Pages that match the prompt and carry evidence are the common factor. The pillar guide to AI search engine optimization covers how AI systems retrieve pages before they write an answer.

What is the Self-Rank Test for best X pages?

The Self-Rank Test is a triage rule for vendor-written "best X" pages that asks two questions: does the page rank your own product first, and does it show first-hand evidence for every product it lists? The answers decide whether to keep, rewrite, merge or retire the page.

The evidence question comes from Google's documentation, which does not mention self-promotional listicles by name. The Write high quality reviews page asks that any page recommending something as "the best" explain why, "with first-hand supporting evidence." For ranked lists it adds: "Ensure there is enough useful content in your ranked lists for them to stand on their own." The helpful content page asks whether content provides "original information, reporting, research, or analysis" and whether it is "primarily made to attract visits from search engines."

The rank-first question reflects the citation data: in Ray's September 2026 sample, AI Overviews left the publisher out of the recommendation 83 percent of the time when they cited its self-ranking list. Google has published no policy, update note or statement about self-ranking lists as of October 2026, and its spam policies do not name them. Their closest rule is scaled content abuse: "many pages are generated for the primary purpose of manipulating search rankings and not helping users." The test does not predict rankings; it flags risk.

The four outcomes of the Self-Rank Test:

• Does not rank your product first, shows evidence: keep it and update dates, prices and features.
• Ranks your product first, shows evidence: rewrite it to rank on stated criteria and say where competitors win.
• Does not rank your product first, no evidence: rewrite it with tests, or merge it into a stronger page on the topic.
• Ranks your product first, no evidence: retire or merge it, and redirect to the closest useful page.

How do you run a Self-Rank Test on your best X pages?

A Self-Rank Test audit takes one spreadsheet: list every best, top and alternatives page, answer the two Self-Rank Test questions, add Search Console clicks and conversions, then keep, rewrite, merge or retire each page and track how AI answers mention the brand afterwards. Plan one review pass per page.

When you check AI answers, test more than the head query. Google's AI features documentation states there are "no additional requirements" to appear in AI Overviews or AI Mode, Google's conversational AI search mode: a page must be indexed and eligible to show with a snippet. Both features "may use a 'query fan-out' technique", issuing related searches across subtopics to build a response, so a comparison page competes on every sub-question, not only the head query.

Common mistakes with self-promotional listicles

The most common mistakes with self-promotional listicles come from treating a list as a ranking hack rather than a buying guide: placing your own product first without proof, publishing many near-identical lists, and reading one study as a Google rule. Each one weakens the page for readers and for the AI systems that cite it.

• Ranking yourself first without evidence. Google's review guidance asks for first-hand evidence behind any "best" pick. Rank on stated criteria and show the tests.
• Publishing a list for every keyword variant. Many thin lists on near-identical queries resemble the pattern Google's spam policies describe as scaled content abuse. Merge them into one strong page per topic.
• Deleting pages in a panic. The 38 percent figure is one consultant's 2026 sample, not a Google announcement. Run the Self-Rank Test and check conversions first.
• Hiding your weaknesses. A comparison that never says where a competitor wins reads as an ad. State the trade-offs, because buyers and AI answers look for them.
• Ignoring independent lists. In Grow and Convert's 2026 data, 85 percent of citations went to third party pages. Work on reviews, analyst coverage and inclusion in independent roundups.
• Letting facts go stale. Prices, plans and features in lists date fast. Show a last updated date and review the page on a schedule. The Google AI content guidelines also ask for a manual fact check of AI-drafted pages before publishing.