Self-Ranked Best-of Lists Barely Move AI Recommendations

You have seen the pages. A competitor publishes "the best web designers in town" or "the top 10 tools for your industry," and the company at number one is the company that wrote it. Then this week a study landed saying those pages help the author get recommended by AI assistants. Now you are wondering whether you need one too, whether Google or your customers would hold it against you, and what else gets a business named when someone asks ChatGPT or Gemini who to hire. On self-ranking best of lists and AI, the short answer is that the measured lift is small, it comes from one vendor's study, and it only applies once an assistant has already chosen to cite the page. Another analysis of Google's AI answers found cited self-promoters left out of the recommendation more often than not. You can check what is actually driving your competitors' AI mentions in about ten minutes, and that check will usually point you somewhere more useful than a list with your own name on top.

Get a Clear Read on What the New Study Found

Scrunch published "What happens when you crown yourself #1: Inside thousands of listicles and their impact on AI answers" on September 24, 2026. Scrunch sells AI visibility monitoring, so this is one vendor's research about the market it sells into. Read it as a useful data point with limits, not as a rule.

What Scrunch measured

The study looked at approximately 10,000 URLs that AI assistants cited between May 17 and June 30, 2026, across ChatGPT, Claude, Gemini, Microsoft Copilot, Perplexity, Google AI Overviews and Google AI Mode. Of nearly 8,000 pages it could classify, comparison and evaluation pages made up 33%. Approximately 30% of those were self-promoting listicles, which Scrunch defines as pages where a company ranks its own product number one.

For the main comparison, Scrunch set 818 self-ranking pages against 1,033 neutral third-party comparison pages and labeled 29,382 answers as recommend, neutral or caution. When a self-promoting listicle was cited, the author brand's recommendation rate rose "from about 4% to roughly 7%," in the Scrunch study's own words. Brand mentions rose from approximately 19% to 39%.

What the study did not show

Scrunch calls the lift "real, if modest," and says a larger study is needed to confirm the headline size. The study measured what happened after an assistant had already cited the list. It did not show that publishing a self-ranked list gets it cited in the first place, and it did not set a rate that applies to every business or every assistant.

PPC Land's report on the Scrunch findings adds more limits from the same data. When a self-promoting list was cited, a competing brand got the recommendation instead of the author 24.3% of the time. The causal findings rest on about 6% of the eligible data, and the mention effect had not yet been through Scrunch's own falsification check. The assistants also behaved differently: Claude and Gemini were the most willing to recommend a list's author, while Copilot rarely did.

See Why One Number Is Not a Plan

The Scrunch figure is getting passed around on its own, but it is not the only measurement. SEO consultant Lily Ray published her own analysis on June 17, 2026. She tracked 100 business-to-business "best [category]" searches in Google AI Overviews on three dates, April 15, May 15 and June 8, 2026. When a brand's own self-promoting list was cited, that brand was left out of the recommendation in 224 of 323 cases, about 69%. Her analysis also found that established brands with strong authority still tended to benefit, while smaller brands were the ones at risk of helping competitors instead.

The two findings do not cancel each other out. They measured different things, on different assistants, over different samples. Put side by side, they show why a single percentage is a weak reason to publish anything.

Read together, the evidence says a self-ranked list sometimes helps its author a little, sometimes hands the recommendation to a rival, and depends heavily on which assistant is answering and how well known the author already is. That is not a tactic to copy. It is a reason to find out what is actually getting your competitors named before you spend a week writing anything.

Keep Clear of Trouble With Google and Your Customers

Google has not published anything that says a company should rank itself first to earn AI recommendations. It has published several rules that apply to these pages, and they are worth knowing before you publish or keep one.

Google's stated position

Google's reviews system guidance covers pages that analyze, research or compare products and services, and it says the system is built to reward "insightful analysis and original research" over thin summaries. A comparison page is judged by that standard.

Google's review snippet documentation says review snippets are not shown for "self-serving reviews," meaning reviews about the business that controls the page. Review markup on a list that ranks your own company will not earn you star ratings in search.

Google's spam policies prohibit content created mainly to manipulate rankings. A self-ranked comparison page is not automatically spam, and Google does not penalize every page that lists its own business first. The risk comes from lists that are thin, exaggerated or mass-produced, which can be demoted or treated as spam. The timing is worth noting too: PPC Land reported that Scrunch's study appeared the same day Google released its September 2026 spam update, the fourth of the year, with a rollout window of up to two weeks. Google has not said that update targets listicles, so do not read it that way. Just avoid publishing anything that would struggle under Google's helpful content guidance.

The customer side

Your customers are the other audience. The Federal Trade Commission's endorsement guidance requires clear disclosure of material connections when readers may not understand the relationship, and claims must be truthful, supported and current. A company ranking itself has an obvious commercial interest, so the page should say so plainly, close to the ranking. A reader who works that out alone may trust everything else on your site a little less.

Find Out What Gets Your Competitors Named in Ten Minutes

Before you write anything, run a quick check. It costs nothing and tells you which sources the assistants are actually using in your category.

Ask ChatGPT, Gemini and Perplexity the same questions, word for word:

• "What are the best [business type] for [customer type] in [market]?"
• "Which [business type] would you recommend for [specific need]?"
• "Compare the best options and cite your sources."
• "What evidence supports each recommendation?"

Write down the date, the assistant, the prompt, the companies named, the order they appear in, every cited link, and whether your business shows up at all. Then sort each cited page into one of these groups: a company's own page, an independent review platform, a directory, a news article, an industry publication, a social or community page, or a competitor's comparison list.

That sorting is the useful part. If your competitors are being named through review platforms, directories or press coverage, another self-ranked list on your own site will not fill that gap. If they are being named through their own clear service pages, that is where your work should start. Keep in mind this is an observation, not an experiment. Answers change with wording, location, account, model version and time, so repeat the same prompts every few weeks instead of trusting a single answer.

Earn AI Mentions From the Sources Assistants Already Use

Published studies of AI citations point to a mix of first-party and independent sources, and they point away from any single page type as a shortcut.

Yext's 2025 analysis examined 6.8 million AI citations from 1.6 million queries across ChatGPT, Gemini and Perplexity, collected July 1 through August 31, 2025. First-party websites and local pages made up nearly 60% of citations for unbranded objective queries. Listings, reviews and social content were also important citation sources, with the mix varying by industry.

A 2025 Seer Interactive study of 800,000 AI responses across ChatGPT, Google AI Mode, Gemini and Perplexity looked at review platforms. It found Trustpilot appearing often in some platform and query groups, and citation patterns that varied a lot from one assistant to the next.

None of this guarantees a mention. No published guidance from Google or any assistant promises that a particular signal will produce a citation, and structured data helps search engines understand a page without making a self-claim more believable. What these sources do have in common is that someone other than you, or a plain fact about your business, is doing the talking. For the broader mechanics, our guide on how to get cited by ChatGPT covers the first-party side in more detail.

Write a Fair Comparison Page if You Still Want One

A comparison page that includes your business can still help buyers, as long as it is written for them rather than for the assistant. If your ten-minute check shows buyers in your category compare options this way, a fair page is a reasonable thing to publish. These are the rules to follow.

Disclose and date it

Say near the top, close to the ranking, that you sell one of the services on the list. State the date you checked the information, the area it covers, and who the page is for. Mention any affiliate links, paid placements or incentives separately, as the FTC's guidance requires.

Use criteria a reader can check

Pick criteria that someone else could verify, such as pricing structure, service area, support model, contract terms, accessibility, integrations or documented capabilities. Apply them the same way to everyone, including yourself, and explain any exception.

Treat competitors fairly

Include competitors because they genuinely fit the category, not because they make you look good. Say where you are strong and where another provider is a better fit for a certain kind of customer. A page that never admits a weakness reads as an ad, to people and to Google's reviews system alike.

Keep it current

Give the page an owner and a review date. Prices, services, reviews and competitors change, and an out-of-date comparison is a liability rather than a help.

These are the failures to avoid, in rough order of how likely they are, based on Google's guidance and the FTC's rules rather than a measured study: unsupported "best" claims, vague criteria that bend toward the publisher, missing disclosure, competitors used as weak foils, thin or repetitive content, expecting the page alone to earn citations, accidentally giving competitors useful mentions, and letting the page go stale.

Measure Results That Reach Your Phone

A mention is not a recommendation, and a recommendation is not a customer. Scrunch's own data shows the gap: mentions roughly doubled when a self-ranked list was cited, while recommendations moved by about three points. Counting mentions alone can make a page look like it is working when nothing reaches your inbox.

Track what matters to the business: qualified inquiries, searches for your business name, referral visits from AI assistants and other sites, and sales you can tie back to a source. Repeat your ten-minute check on a schedule so you can see whether your visibility is changing across several assistants, not just one.

Be wary of anyone who promises a specific recommendation rate, citation rate, ranking or sales result from AI assistants, including us. The evidence does not support that kind of promise. If your business is missing accurate listings, genuine reviews, clear service pages or any outside coverage, fixing those will likely do more than any new page, and some of that work you can do yourself for free.

Frequently Asked Questions About self-ranking best of lists ai

Does ranking my business first in a best-of list make AI recommend me?

Not reliably. Scrunch's vendor study found the author's recommendation rate rose from about 4% to roughly 7% when an assistant cited the list, but it did not show that publishing the list causes citations or recommendations. A separate June 2026 analysis of Google AI Overviews found authors left out of the recommendation in 69% of cited cases.

Will Google penalize a page that ranks my own business first?

Not automatically. Google judges usefulness and manipulation. Self-serving reviews are not eligible for review snippets, and thin, exaggerated or mass-produced lists can be demoted or treated as spam.

What is the difference between being mentioned and being recommended by AI?

A mention names your business. A recommendation actively presents it as an option to choose. In Scrunch's data, mentions rose far more than recommendations when a self-ranked list was cited.

What else gets a business mentioned by AI assistants?

Studies from Yext and Seer Interactive found assistants citing a mix of first-party websites and local pages, listings, reviews and other outside sources, with the mix varying by industry and assistant. Clear service pages, accurate listings, genuine reviews and independent coverage are the places to start.

Should I list competitors on my own website?

Only on a comparison page that genuinely helps buyers. Disclose that you sell one of the options, use criteria a reader can check, and say where a competitor is the better fit.

How can I check whether AI assistants recommend my business?

Ask ChatGPT, Gemini and Perplexity the same buyer questions, record the companies and cited sources, sort each source by type, and repeat the check every few weeks. For the basics of how these engines choose sources, see our explainer on generative engine optimization.

Moving Forward

The new study gives self-ranked lists a small, conditional lift, measured by a vendor, and only after an assistant had already chosen to cite the page. Another analysis of Google's AI answers found the opposite result for most cited authors. Google does not reward self-serving reviews with snippets, its spam policies cover content made to manipulate rankings, and the FTC expects the commercial relationship to be obvious to readers.

The more dependable route is slower but it lasts: clear service pages, accurate listings, real reviews and mentions from people who are not you. Run the ten-minute check, find out which of those sources are naming your competitors, and fix the gaps you find. If you still want a comparison page after that, write it for the buyer, date it and disclose it.

If you want help with that check, Web Leveling can run the same buyer questions across several assistants, sort the sources they cite and show you which gaps matter for your category. Our generative engine optimization work covers your own pages, listings and the outside sources assistants draw on, and we will not promise a recommendation rate that no one can deliver. We work with small and medium businesses across the country and overseas. Tell us who the assistants are naming instead of you, and we will help you work out why.

Terms