Listicles and AI Citations: Do Self-Promotional Lists Still Work?

Yes, self-promotional "best X" listicles still get cited by AI search engines: in Peec AI's study of 232,000 citations in the software niche, about 1 in 10 came from lists where the publisher ranked its own product, with no sign of the engines correcting it over 12 weeks. But being cited is not the same as being recommended. When Lily Ray checked 100 "best software" queries in Google AI Overviews, brands whose own listicle was cited were left out of the recommendation 69% of the time, and the answer named other brands instead.

This article puts the seven published studies on listicles and AI citations side by side, with sample sizes and caveats, adds what we see in AI Rank Lab's own citation data, and turns it into practical rules: when a self-published list is legitimate, what Google's policies and the engines penalize, a checklist for writing a list that deserves a citation, and how to get onto other people's lists honestly. We publish "best tools" lists ourselves, so we say how we handle that too. Studies and vendor pages checked October 24, 2026.

Do self-promotional listicles still get cited by AI search engines?

They do, at a stable rate, on every major engine measured so far. Peec AI's self-promotional listicles analysis (Tom Wells, March 2026) tracked the top 1,000 retrieved listicles per platform each week for 12 weeks, from December 2025 to February 2026, across ChatGPT, Google AI Mode, Perplexity, Microsoft Copilot, Gemini and Google AI Overviews. A list counted as self-promotional when the publishing domain's own brand appeared in it as a recommended product. The week-on-week share hovered around 10% of citations, and the author's conclusion was blunt: AI search "hasn't found a way to actively filter self-promotional content (yet)."

Two qualifiers matter. First, this is software reviews, the niche Peec picked because it is full of self-promotion, so the share will be lower in most other categories. Second, citation share is the wrong success metric on its own. The more recent studies, covered below, show that a cited list often helps the brands it mentions more than the brand that wrote it.

What share of AI citations come from listicles?

It depends on what you count, which is why published figures range from about 11% to over 40%. Ahrefs' Glen Allsopp counted all "best X" blog lists and found they made up 43.8% of the source page types in ChatGPT answers for 750 prompts about software, products and agencies. Seer Interactive used a narrower URL rule (any URL containing "best" or "top") and measured listicles at 17.2% of ChatGPT citations in December 2025 and 15.5% in January 2026. Peec counted only self-promotional lists and landed near 11%. These numbers do not contradict each other; they measure different slices of the same pile.

What the studies found: listicles and AI citations, 2025-2026

Seven studies with published methods cover this question. All are by companies that sell SEO or AI visibility tools or services, including two by Ahrefs and three by Peec, so read them as informed observational data, not neutral science. None is a controlled experiment across categories.

Where the studies agree, and where they do not

They agree on three things. Listicles are heavily retrieved for "best" and "top" prompts. ChatGPT leans on self-promotional lists less than Google AI Mode and Perplexity do. And engines treat a list as a source of candidates rather than as a verdict: they read who is on it and then decide.

They disagree on direction of travel. Peec saw no systemic correction through February 2026; Seer saw ChatGPT pull back from listicles in January 2026; Peec's own later notes describe a further ChatGPT shift toward product pages in August 2026. The most likely reading is that ChatGPT is tightening while Google's AI surfaces and Perplexity lag, but no study yet covers the whole of 2026 across all engines.

There is also a real counter-argument on the Google side. Grow and Convert's response to the listicle crackdown claims (updated May 26, 2026) points out that on the blogs Ray first reported (blog traffic down 29% to 85% where a figure was given), self-promotional listicles were only 0.6% to 9.6% of posts. Their argument: Google is demoting low-quality content in general, not lists that include the author. That is worth keeping in mind before deleting every comparison page you own.

What we see in AI Rank Lab's own citation data

In the prompts tracked in AI Rank Lab, the third-party sites that get cited most are mostly software vendors, the same kinds of companies that publish "best tools" lists about their own category. We counted cited domains across successful prompt runs stored in AI Rank Lab between February and October 2026: 1,676 runs with at least one citation, for prompts tracked by 14 domains, across ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews and Google AI Mode. About 56% of those runs come from software and marketing-tool brands (the rest from real estate, energy, finance and other sectors), so the sample leans toward "best tool for X" prompts.

We then checked, on October 24, 2026, what three of the most-cited vendors publish on this topic. Conductor's "10 Best Enterprise AEO Tools in 2026" (updated September 10, 2026), SE Ranking's "10 best AI visibility tracking tools" for its SE Visible product (August 20, 2026) and Profound's "19 best answer engine optimization platforms" (July 28, 2026) each rank the publisher's own product first. All three describe competitors' strengths. None carries an explicit statement that the publisher has a commercial interest in the ranking that we could find, beyond the page living on the vendor's own site.

Three caveats keep this honest. Our dataset records cited domains, not which page on each domain was cited, so we cannot say these citations went to those listicles specifically. Fourteen tracked domains is a small, self-selected sample of one industry; your category will look different. And the sources move a lot: across 972 repeated runs of 42 prompt-engine pairs (February to October 2026), the median overlap of cited domains between consecutive runs of the same prompt was 0.41, and the median share of cited domains present in every run was 3.4%. We cover that variance in our AI answer consistency study; it matches Ahrefs' finding that its pages were cited on only about one in three eligible days.

We publish self-promotional lists too

AI Rank Lab publishes "best tools" and "alternatives" articles on this blog, and we put AI Rank Lab first in them. Our house rule for those posts is a disclosure line at the top of the list ("Disclosure: AI Rank Lab is our product. We list it first because this is our site, and we've kept the same criteria for every tool below."), the same criteria applied to every tool, and a "where this tool is the better choice" note for each major competitor. Some older list posts on this blog predate that rule and do not carry the line yet. The checklist later in this article is the standard we hold our own lists to, and Ray's data applies to us as much as anyone: an engine may well cite our list and recommend a competitor from it.

Why do AI engines cite "best X" listicles so often?

Because a good listicle answers the exact shape of a selection query. When someone asks an assistant for the best tool for a job, the engine expands the prompt into several hidden searches, a process Google defines in its guide to optimizing for generative AI features as query fan-out: "a set of concurrent, related queries generated by the model to request more information." Peec's research notes that "best" is the word ChatGPT adds most often to those hidden searches, and a "best X" list is the page built around that word.

A strong list also does several jobs at once, which helps it survive reranking:

• It names many entities in one place. The engine needs candidates; one page gives it eight or ten.
• It covers comparison attributes. Price, use case, integrations and limits map to the "X vs Y" and "X for Y" sub-queries in a fan-out.
• It has extractable passages. A short "best for" line under each heading is easy to lift into an answer.
• It is usually fresh. Ahrefs found 79.1% of the lists ChatGPT cited had been updated in 2025, and 26% in the two months before the study.

None of this is a bias for the list format as such. The same logic is why definitions win "what is X" prompts and numbered steps win "how to" prompts. If you want the wider picture of which source types each engine prefers, see what sources LLMs cite; this article stays on listicles.

Does being cited in your own listicle mean the AI recommends you?

Often not. The clearest evidence is Ray's June 2026 dataset: across 100 B2B software queries in Google AI Overviews, a brand's own "best" list was cited but the brand was not recommended in 224 of 323 cases, or 69%. The engine used the page as research and recommended other brands; in Ray's reading, the ones that won were usually category leaders with strong third-party mentions.

Ahrefs reached a similar result from the other direction. In its own experiment, 43% of answers that cited one of its conference pages never mentioned the conference. The split by brand maturity is the most useful finding in the whole body of research:

• New or niche brand. For the new Ahrefs Evolve conference, 82% of the 72 new mentions it gained appeared in answers that cited the self-published pages. When the engine has no prior knowledge of you, your own page can introduce you.
• Established brand. For Ahrefs Brand Radar, only 6% of new mentions came from answers citing those pages; 94% came from third-party content. When the engine already knows you, your own claim adds little.

This is the same distinction we draw in brand mentions vs citations: a citation is a link to a source, a recommendation is your name in the answer. Ray's conclusion is that recommendations track how often a brand is mentioned and linked across the web, not what it says about itself. Measure both, separately.

Does Google penalize self-promotional listicles?

Google has no policy against listing your own product, but it has three policies that a careless listicle program can trip, and it has said publicly that it is working on the problem. On April 14, 2026, Search Engine Roundtable reported a statement Google gave The Verge: spokesperson Jennifer Kutz said the company applies "robust protections against common forms of manipulation in search and Gemini" and is aware of low-quality listicle content.

Which Google policies apply to "best" lists?

• Manipulating AI answers is spam. Google's spam policies (last updated August 28, 2026) include attempting to manipulate generative AI responses in Google Search. A page written mainly to steer AI answers, rather than to help a buyer choose, is in scope.
• Scaled content abuse. Many near-identical "best X for Y" pages generated mainly to rank, especially with AI tools and no added value, are the textbook case.
• Site reputation abuse. If you host third-party "best of" pages on your domain to borrow its ranking signals, without real editorial oversight, that is a separate violation.
• The reviews guidance sets the quality bar. Google's guidance on writing high-quality reviews (updated December 10, 2025) asks you to "cover comparable things to consider," back up any "best" claim with first-hand evidence, and "ensure there is enough useful content in your ranked lists for them to stand on their own."

On enforcement, the evidence is observational. Ray reported that dozens of sites saw organic drops around January 20, 2026, and that the affected sites shared a pattern of self-ranking lists at scale. Google has not confirmed any update aimed at listicles. Google's AI Overviews and AI Mode retrieve their sources from the Google index, so a sitewide demotion there also shrinks what those features can cite from you.

What are the trust and legal risks?

Readers increasingly recognize a list that crowns its author, and Ray notes that Google has started showing disclaimers in AI Overviews for categories saturated with self-promotion. In the US, the FTC's endorsement guides FAQ says material connections must be disclosed clearly and conspicuously, that a disclosure hidden behind a link is not clear and conspicuous, and that "the closer the disclosure is to your recommendation, the better." That guidance is written for endorsements and affiliate relationships, but the principle carries over cleanly: if you sell one of the products you rank, say so where the reader sees the ranking.

When is a self-published list a reasonable bet?

How do you write a listicle that deserves to be cited?

Write it so it would still be useful if your product were removed from it. That single test covers most of what Google's reviews guidance, the FTC and the engines reward. Here is the checklist we use, with the evidence behind each line.

In practice, write it in this order:

1. Pick a query you genuinely fit. Ahrefs' new conference appeared in 66.4% of "best SEO conferences 2026" answers but only 15.8% of "best marketing conferences 2026" answers. Fit beats ambition.
2. Write the criteria before the entries. Decide what matters to the buyer, then score everyone, including yourself, against it.
3. Research competitors properly. Check their pricing and docs on the day; give each a "best for" line you would accept if they wrote it about you.
4. Add the disclosure and the method. Say who wrote it, what you tested and when.
5. Run the swap test. If a competitor sat in your #1 spot, would the page still be accurate and useful? If not, rewrite it.
6. Maintain it. Recheck prices and features on a schedule; a stale list hurts trust faster than any algorithm.

For the wider trust signals that sit around a page like this (authorship, sourcing, transparency), see our guide to E-E-A-T for AI search.

How do you get included in third-party listicles legitimately?

Start with the lists the engines already cite for your prompts, not every list that exists. Peec's rank-effect study found the value of placements plateaus after a few repeated appearances, and that the lift comes from frequently retrieved third-party lists, so a short target list beats a long outreach campaign.

1. Run your category prompts and collect the cited URLs. Use the prompts buyers actually ask ("best X for Y", "X alternatives"), on each engine separately, more than once. Answers vary run to run, so one check is not enough.
2. Keep the lists that recur. Filter for URLs that show up across several prompts and engines. Those are the pages doing the work.
3. Find the gap. Note which lists name your competitors and not you. This is a source gap, and our AI content gap analysis guide shows how to score it alongside prompt and topic gaps.
4. Check who controls the page. Confirm the author still edits it and that it is editorial, not a paid directory.
5. Give the editor evidence, not adjectives. Peec recommends supplying feature pages, reviews, integration guides and transparent pricing. A short, factual description they can paste helps.
6. Build review-platform presence. G2 was the only review platform in our top 20 most-cited domains, with 140 citations. Real customer reviews there feed the same retrieval.
7. If you pay, make sure it is disclosed. Paid or sponsored placement exists, and the Peec study notes it as an option. An undisclosed paid ranking is exactly the kind of material connection the FTC guidance says must be disclosed.

Tools that show which listicles AI engines cite

Disclosure: AI Rank Lab is our product. We list it first because this is our site, and we've kept the same criteria for every tool below. The criterion here is simple: does the tool show you the URLs AI engines cite for your own prompts, so you can find the lists worth pitching?

• AI Rank Lab. Runs the prompts you choose on ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews and AI Mode and stores every run; citation analytics shows which domains and URLs are cited for those prompts and your competitors' citation share. It does not run prompts on a schedule by itself: you re-run them when you want a fresh read, and only prompts you opt into the Action Plan citation tracker are re-checked weekly.
• Ahrefs Brand Radar. The data source behind both Ray's AI Overviews study and Ahrefs' own experiment, with a database of 150M+ prompts per Ahrefs' December 2025 study; a better choice if you want to see cited pages across prompts you have not written yourself.
• Peec AI. An AI search monitoring platform whose research team published three of the studies above; a strong choice if you want source and listicle analysis framed the way those studies frame it.

How do you measure whether a listicle is helping?

Track whether you are recommended, not just whether a list is cited. Peec's rank-effect study measures three separate outcomes, and they are a good template: mention probability (how often you are named), answer position (how early), and mention count (how often within one answer). Add a fourth for your own lists: when your page is cited, how often are you the brand recommended?

Measure against a baseline taken before the change, on the same prompts and engines, across several runs, because sources rotate heavily from run to run. On Google, the Search Console Generative AI performance reports show how your pages perform in AI features, which is the one place Google itself reports that data. To do this in AI Rank Lab, record a baseline with brand visibility runs before you publish or pitch, then re-run the same prompts afterwards and compare the stored runs.

Method and limitations

Study figures come from each publisher's own write-up, read on October 24, 2026; where a figure appeared only in secondary coverage, we left it out. Several studies are by vendors with a commercial interest in AI visibility (including Peec AI, Ahrefs and us), and none is a randomized test across categories. AI Rank Lab figures are aggregated and anonymized: cited-domain counts from 1,676 successful prompt runs with citations for prompts tracked by 14 domains (February to October 2026), and run-to-run stability from 972 runs of 42 prompt-engine pairs with at least five runs each. We name only well-known public vendors and publishers from that list. Vendor list pages were checked on October 24, 2026 and may have changed since.

If you want to see which listicles AI engines cite for your own category prompts, and which of them name your competitors and not you, you can try AI Rank Lab: signup includes 100 free credits, no credit card needed.