Is Google Cracking Down on 'Best Of' Comparison Articles?
| No.1 Claims, Legal Risks, and AI Search
Hi, I'm Tsubasa.
I write about things I learn on the job as an EC (e-commerce) operator, and comparison articles — the "best X services" or "top Y tools" format — tend to attract the most traffic on this blog. I've written comparison pieces on retouch companies, SEO agencies, product listing services, freelance matching platforms, and more. It's satisfying when they rank well and get read.
But writing comparison articles myself has given me a particular vantage point. In May 2026, Google explicitly warned against manufactured brand mentions. Companies that claim "No. 1" without evidence face legal risk under advertising regulations. And AI search still cannot filter out these problems. The comparison article format itself isn't the issue — but understanding the structural problems beneath it matters.
Google declared manufactured brand mentions equivalent to spam. Meanwhile, an Ahrefs study (December 2025, analyzing 26,283 URLs) found that 43.8% of pages cited by ChatGPT were "best of" list articles, and 28% of those had zero organic traffic from Google. In Japan, 13 enforcement actions were issued in fiscal year 2023 alone under the Act against Unjustifiable Premiums and Misleading Representations for unsubstantiated No. 1 claims. This article examines the current state of comparison articles from three perspectives: Google's official position, AI search behavior, and legal regulation.
Contents
- Google Declared Inauthentic Brand Mentions as Spam (May 2026)
- Are Paid Business Matching Sites' "Best Of" Articles Going to Be Flagged as Spam?
- Unsubstantiated "No. 1" and "Lowest Price" Claims Are Illegal Under Advertising Law
- AI Search Cannot Filter Comparison Article Problems Yet
- Google AI Overviews Has Started Separating "Citation" from "Recommendation"
- What to Check When Reading Comparison Articles
- FAQ
- Summary
Google Declared Inauthentic Brand Mentions as Spam (May 2026)
On May 15, 2026, Google published an official guide titled "Optimizing your website for generative AI features on Google Search." Within the "Mythbusting" section, a passage on "Seeking inauthentic mentions" stands out.
Google stated that while generative AI features can surface what's being said about products and services across blogs, videos, and forums, "seeking inauthentic mentions across the web may seem helpful on the surface, but it really isn't that useful. Our core ranking systems focus on quality content, and other systems are designed to block spam. Our generative AI features fully depend on both."
However, the reality has not caught up with this official stance. As of now, manufactured mentions and self-promotional "best of" articles still appear in both regular and AI search results. It would be inaccurate to say Google's spam systems are effectively blocking them. Google declaring something "not useful" and Google's systems actually preventing it from working are two different things.
This is especially notable because SEO companies themselves — the very industry most directly affected by this policy — compete heavily in publishing "best of" comparison articles. Users rely on these articles because there's often no better alternative for comparing service providers. Unless Google offers a substitute, eliminating all such content entirely would actually harm users.
On the same day, at Search Central Live Sydney 2026, Google's Gary Illyes went further. When an AI platform promoted a feature that "automates the purchase of mentions to place brand references within AI responses," Illyes stated this would be treated as paid-link-level spam (Search Engine Roundtable, May 2026).
Google also updated the opening paragraph of its search spam policies to include "attempts to manipulate generative AI responses in Google Search" in the definition of spam. This made any deliberate attempt to distort AI Overviews or AI Mode output a clear policy violation.
An important translation note: Japanese SEO expert Kenichi Suzuki pointed out (海外SEO情報ブログ, May 2026) that Google's Japanese documentation translates "inauthentic" as "inaccurate" (不正確な). The intended meaning is "manufactured" or "staged" — it's about intent, not factual accuracy. Reading only the Japanese version obscures how serious this warning is.
Note that what Google is targeting here is the act of paying or incentivizing third-party sites to mention your brand. Mentioning your own company on your own website is not covered. Writing about your own service on your own blog is standard marketing and perfectly fine.
Are Paid Business Matching Sites' "Best Of" Articles Going to Be Flagged as Spam?
Behind Google's spam declaration lies a structural problem with the flood of "best X" articles on the web. In the Japanese market particularly, B2B business matching platforms produce enormous volumes of recommendation articles.
These platforms publish articles like "Top 3 Retouch Companies" or "10 Best System Development Firms" to dominate search result real estate. The format gives the impression that companies were carefully selected from across the entire market.
But business matching sites generate revenue through listing fees and referral commissions from the companies they feature. Every company in a "best X" article has potentially paid the platform in some form. If companies could get listed for free, why would anyone pay? The business model wouldn't work.
Looking at these articles closely, well-known industry players are often listed alongside companies that barely appear anywhere else on the web. Search for the featured company names outside of that matching platform, and the difference becomes apparent.
The matching site's homepage typically makes it clear that it's a paid service. Nothing is hidden. But when a reader lands on a "best X" article directly from search results, they may not realize this and instead interpret it as "a neutral third party surveyed the industry and picked the best options." The gap between the article's format and its actual selection criteria is the core of this problem.
Whether Google's algorithm will eventually act against paid matching site articles is unknown — it depends on future algorithm updates. But the risk is not zero.
Unsubstantiated "No. 1" and "Lowest Price" Claims Are Illegal Under Advertising Law
Up to this point, the discussion has been about Google policy and matching site structures. Here, a separate legal issue needs to be addressed.
Writing about your own service at the top of a comparison article on your own blog is not inherently problematic. Promoting your company on your own site falls within standard marketing.
The problem arises when "No. 1," "best quality," or "industry's lowest price" is stated without objective evidence. In Japan, this can constitute a violation of the Act against Unjustifiable Premiums and Misleading Representations (Article 5) — either as a superior-quality misrepresentation or an advantageous-terms misrepresentation.
Japan's Consumer Affairs Agency has laid out four requirements for No. 1 claims to be considered "based on reasonable grounds" (Consumer Affairs Agency, "Survey Report on No. 1 Claims," September 2024): the comparison products must be appropriately selected, the survey respondents must be fairly chosen, the survey methodology must be objective, and the claim must accurately reflect the survey results. No. 1 claims that fail to meet these four requirements are subject to cease-and-desist orders and surcharge payment orders.
Enforcement has intensified. In fiscal year 2023 alone, 13 cease-and-desist orders were issued for No. 1 claims. In September 2024, the Consumer Affairs Agency published a comprehensive survey report and further strengthened monitoring (Consumer Affairs Agency, Fiscal Year 2024 Press Releases). Surcharges are set at 3% of the relevant product's revenue, increasing to 4.5% for repeat violations.
Even heavier penalties have been imposed. In March 2024, the Consumer Affairs Agency issued a three-month business suspension order against an EC operator selling diet coffee under the Specified Commercial Transactions Act (Consumer Affairs Agency, March 15, 2024). The company had displayed "10 Crowns Achieved" and "No. 1 Diet Drink Popular Among Women," but the underlying survey only targeted members of a commissioned research firm and was not considered fair or objective. Combined with failure to display six required items on the final purchase confirmation screen for subscription products, the company received both cease-and-desist and business suspension orders. A personal business prohibition order was also issued against the CEO.
While these examples are from Japan, similar regulations exist in many jurisdictions. The U.S. Federal Trade Commission (FTC), the UK's Advertising Standards Authority (ASA), and the EU's Unfair Commercial Practices Directive all restrict unsubstantiated superiority claims in advertising.
This matters beyond articles. Whether on a homepage headline, an about page, or a pricing page — the location doesn't matter. Unsubstantiated claims carry the same legal risk wherever they appear.
And here's the crucial point: neither Google's search algorithm nor any major AI system — ChatGPT, Gemini, or others — currently has a mechanism to detect and demote these illegal claims. If a site states "No. 1" or "lowest price" without evidence, AI takes it at face value and may present it as a recommendation to users. This does not serve users' interests.
AI Search Cannot Filter Comparison Article Problems Yet
Ahrefs published large-scale verification data in December 2025 on how AI search handles comparison articles (Ahrefs Blog, December 2025). Analyzing 26,283 source URLs cited by ChatGPT, several serious findings emerged.
First, 43.8% of pages cited by ChatGPT were "best of" list articles. Even in niche categories like retouch services, SEO agencies, or product listing outsourcing, AI doesn't evaluate companies from scratch — it pulls from existing list articles on the web.
More concerning, 28% of the cited pages had zero organic traffic from Google. Pages nobody finds through Google search are being used as source material for AI responses. AI has a bias toward citing recently updated content (79.1% of cited list articles were updated within 2025), meaning low-quality articles that merely change the year in the title without updating the substance can still enter AI responses simply by being "new."
This mechanism allows a company to write an article placing itself at No. 1, and AI may reflect that directly in its response. Even when AI can detect that the article's author and the recommended company are the same entity, it doesn't discount this and presents it as a neutral recommendation.
Even explicitly prompting AI to "recommend based only on third-party evaluations" doesn't fully solve the problem. In niche industries with limited web content, a company's own blog may be one of AI's only available sources. No matter how carefully users craft their prompts, AI can only work with the articles that exist.
On the other hand, Google's search algorithm is beginning to act. SEO expert Lily Ray (Amsive) found that several SaaS companies that heavily relied on self-promotional list articles lost 30% to 50% of organic traffic after a January 2026 Google algorithm update (Search Engine Land, March 2026). Traffic losses concentrated in the blog subfolders hosting "best of" articles, suggesting Google may be specifically targeting self-promotional list content.
However, AI search has not caught up yet. Considering it took Google roughly a decade from Penguin in 2012 to SpamBrain in 2022 to address link spam, AI-specific countermeasures could similarly take years.
Google AI Overviews Has Started Separating "Citation" from "Recommendation"
The previous section stated that AI search hasn't caught up yet, but the situation shifted in June 2026. At least for Google AI Overviews, a response to self-promotional list articles has begun.
SEO expert Lily Ray (Algorythmic / Amsive) published a study on June 17, 2026 (Lily Ray Substack, June 2026) that quantitatively demonstrates this change. Ray collected Google AI Overview response text and citation sources for 100 B2B "best [category] software" queries across three time points — April, May, and June 2026 — and separately tallied "cited" versus "recommended" brands.
The results were clear. In cases where self-promotional list articles were used as AI Overview citation sources, the self-promoted brand was excluded from the recommendation 69% of the time. Across 100 queries, 74 showed self-promotional list articles being cited while the promoting brand was absent from AI Overview's actual recommendations.
The articles are being cited. Google's system reads them and uses them as information sources. But the "we're No. 1" claim within those articles doesn't translate into AI's recommendation. The system extracts information about competitor brands from the article and recommends those instead.
What this means in practice: a self-promotional list article ends up functioning as a recommendation vote for competitors. In Ray's study, Pylon and Crisp each published list articles naming themselves as the top help desk software. But AI Overview recommended Zendesk, Freshdesk, and Help Scout instead. Pylon's and Crisp's articles were used as citation sources, but the brands recommended were the competitors mentioned within those articles. From the article authors' perspective, they had been boosting their competitors' positions.
Brands that ended up being recommended shared common characteristics: high backlink counts, frequent brand mentions within AI Overviews, and high brand-name visibility in ChatGPT as well. Companies widely discussed by third parties across the web get recommended. Companies that only promote themselves get cited but not recommended.
Ray concluded that instead of tracking "citation count" as an AI search metric, businesses should monitor whether they appear in actual recommendations. As voice search grows, users click on links less and less. Whether AI mentions a brand name in its response text matters more than whether the source article happens to be cited.
I mentioned at the beginning of this article that I've written many comparison articles myself on this blog. However, I've never labeled any specific company as "No. 1." The reasons were legal risk and fairness to readers. Reading Lily Ray's study, it turns out this was also the right approach for AI search dynamics. The more you call yourself No. 1, the more AI extracts your competitors' information from your article and recommends them instead. Self-proclaimed No. 1 list articles are becoming counterproductive — both legally and in AI search.
L Why Calling Yourself the "Best" Could Be Helping Your Competitors Win in AI Search lilyraynyc.substack.comWhat to Check When Reading Comparison Articles
Based on everything covered above, here's what to verify when reading "best of" or comparison articles.
First, check who runs the site. Is it a company's owned media, a paid matching platform, or a personal blog? Don't judge by the article URL alone — look at the site's top page and footer for operator information. That alone reveals a lot about the article's context.
Next, search for the featured company names outside of that site. Use a different search engine or ask an AI. If a company barely appears anywhere except in articles on that one matching platform, the selection criteria may not represent the full industry.
If you see "No. 1," "lowest price," or "best quality" claims, check whether survey methodology and source data are cited. If there's no link to objective survey data and no explanation of the basis, that claim is legally risky self-promotion.
Don't take AI-generated "recommendations" at face value either. As noted above, current AI systems don't discount the fact that the article author and the recommended company are the same entity. When AI recommends a service, open the source article and check who wrote it and for what purpose.
As someone who writes comparison articles myself, I've listed the comparison articles published on this blog in the Japanese version of this article. I invite readers to evaluate the selection criteria of each article for themselves.
FAQ
Q. Is Google trying to ban "best of" comparison articles?
A. Google is not banning the comparison article format itself. What Google targeted in May 2026 was the act of paying or incentivizing third-party blogs and forums to mention your brand (inauthentic mentions). Writing about your own services on your own site, or third parties independently writing comparison articles, is not covered.
Q. Is it a problem to list my own service first in a comparison article on my own blog?
A. That's fine. Promoting your company on your own site is standard marketing. The problem is labeling your service as "No. 1." Placing your service first and declaring it No. 1 are different acts. Stating "No. 1," "best quality," or "lowest price" anywhere on your site without objective survey evidence may violate advertising regulations. In Japan, this falls under the Act against Unjustifiable Premiums and Misleading Representations. Similar rules exist in many other jurisdictions — the U.S. FTC, UK ASA, and EU regulations all restrict unsubstantiated superiority claims.
Q. Will paid business matching sites' recommendation articles disappear from search results?
A. Unknown at this point. Google has declared inauthentic mentions equivalent to spam, but whether its algorithm will specifically act against paid matching sites' articles is up to Google. Given that it took about a decade to fully address link spam, this won't happen overnight — but it won't take a decade either. The timeline is likely measured in years.
Q. Can I trust AI recommendations when I ask "what's the best X"?
A. Current AI search does not discount the fact that the article author and the recommended company are the same entity. According to Ahrefs' study (December 2025), 43.8% of pages cited by ChatGPT were "best of" list articles, and 28% of those had zero Google organic traffic. After receiving AI's recommendation, open the source article and check who wrote it and why.
Q. Have there been actual penalties for No. 1 claims?
A. In Japan, 13 cease-and-desist orders related to "customer satisfaction No. 1" and similar claims were issued in fiscal year 2023 alone. In September 2024, the Consumer Affairs Agency published a survey report and strengthened monitoring. Surcharges are set at 3% of relevant revenue, increasing to 4.5% for repeat violations. In March 2024, a health food EC operator received a three-month business suspension order under the Specified Commercial Transactions Act — heavier than a cease-and-desist — when No. 1 claims combined with subscription display violations.
Q. What should I be careful about when writing comparison articles?
A. When featuring your own service, treat it on equal footing with competitors. Avoid unsubstantiated superiority claims (No. 1, lowest price, best quality). Cite your information sources. Select comparison targets from the same service category. Ensuring readers can understand the article's selection criteria builds long-term trust.
Q. If I write a list article ranking my company first, will AI recommend my competitors instead?
A. According to Lily Ray's June 2026 study, when a self-promotional list article was used as a citation source by Google AI Overviews, the self-promoted brand was excluded from the recommendation 69% of the time. AI uses the article's information but doesn't adopt the self-promotional ranking — instead, it tends to recommend competitor brands mentioned in the article. Your own list article may end up functioning as a recommendation vote for your competitors.
Summary
The comparison article format itself isn't the problem. The issues break down into three areas. First, paying third-party sites to mention your brand is what Google has explicitly declared spam. Second, writing "No. 1," "best," or "lowest price" without evidence violates advertising regulations in many jurisdictions. Third, paid business matching platforms' "best X" articles have a gap between their format and actual selection process. Lily Ray's June 2026 study quantitatively showed that Google AI Overviews has started separating citation from recommendation. Self-promotional list articles get cited as sources, but the promoting brand is excluded from the recommendation 69% of the time — and competitors mentioned within the article get recommended instead. ChatGPT cites "best of" list articles in 43.8% of its responses, and 28% of those have zero Google organic traffic (Ahrefs study, December 2025). Meanwhile, many industries — starting with SEO companies — compete in publishing recommendation articles, and users rely on them for choosing service providers. Unless Google offers an alternative, eliminating all such content would actually harm users. That's precisely why readers need to develop the habit of checking the background behind comparison articles.
🇯🇵 この記事の日本語版を読む
よく読まれている記事
