Why Self-Ranking Listicles Are Losing Ground in AI Overviews

What changed for self-ranking listicles in AI Overviews?

The measurable change is that Google AI Overviews cited fewer self-ranking listicles in Lily Ray's September 2026 research than in her June measurement. Ray reported that, across 100 B2B software queries, AI Overviews cited 38% fewer vendor-written best-of pages in September. She also found that the brand placing itself first was left out of the recommendation in 83% of AI answers that cited the page, up from 69% in June. PPC Land, 2026-10-05

That does not establish a universal penalty for every comparison page. The sample covers one vertical, one search surface, and the published account does not provide a full methodology for matching the June and September query sets. It does establish a useful editorial warning: a citation is not the same thing as a recommendation. An AI Overview can use a listicle as background, then name a competitor rather than the company that wrote the list.

The distinction matters because self-ranking listicles were designed around a simpler assumption. A vendor creates a page for a query such as best project management software, ranks itself at number one, and hopes search engines or answer systems repeat that placement. The September result suggests the system can separate the page from the conclusion the publisher wants it to reach. The page may contribute information while the self-serving claim fails to become the answer.

Google's own guidance gives the durable context. Google says there are no extra technical requirements or special optimization needed to appear as a supporting link in AI Overviews or AI Mode. Eligible pages must be indexed and eligible to appear with a Search snippet, while existing SEO practices remain relevant. Google Search Central, 2025-12-10

1. June 2026: Ray reported that the self-ranking brand was omitted in 69% of relevant AI Overview recommendations.
2. September 2026: Ray reported omission in 83% of relevant AI Overview recommendations.
3. September 2026: Ray reported 38% fewer citations of vendor-written self-ranking pages versus June.
4. The evidence is directional, not a guaranteed outcome for every listicle.

Why can a cited self-ranking page still fail to recommend its publisher?

A cited self-ranking page can fail to recommend its publisher because AI Overviews assemble an answer from supporting links rather than simply copying one page's order. Google says AI has may use query fan-out, issuing related searches across subtopics and data sources while identifying supporting web pages. That process creates room for an answer to use a vendor's category framing while relying on other information for the final recommendation. Google Search Central, 2025-12-10

This is the weakness in an article whose central proof is its own ranking. A reader sees a vendor say that it is number one, but the claim has no independent force unless the page also shows how products were selected, what was evaluated, which limitations matter, and where the supporting evidence came from. An answer system has the same problem. A self-placement is a claim from an interested party, not independent confirmation.

The PPC Land report adds an important caution. It describes conflicting findings from other studies: one study of organic traffic reported gains for many vendor-authored list pages, while another multi-assistant study found a smaller recommendation lift when self-ranking pages were cited. Those results measure different systems and outcomes. Organic traffic, AI Overview citations, and recommendation rate are not interchangeable metrics. PPC Land, 2026-10-05

That is why citation work needs clean measurement. Track whether a page is indexed and visible, whether it appears as a source, and whether the resulting answer recommends the brand. Omnicite calls the percentage of relevant AI answers that cite a brand Citation Share. Citation Share answers a different question from whether the answer selects the brand as the best fit.

1. A source link can supply context without supplying the answer's final recommendation.
2. A vendor's own ranking is an interested claim unless it is supported by transparent evidence.
3. Organic search traffic does not prove AI Overview citation performance.
4. Citation Share measures cited answers, not merely page impressions or classic rankings.

Who does this shift affect most?

This shift affects publishers whose comparison strategy depends on ranking themselves first without giving readers a credible way to evaluate the ranking. B2B software companies are the clearest group in Ray's sample because the 100 queries focused on best B2B software categories. Agencies that create large batches of near-identical alternatives pages also face exposure when the ranking logic is repetitive and the evidence is thin. PPC Land, 2026-10-05

It also affects editorial teams that have treated listicle production as a shortcut to authority. A long list is not automatically useful because it includes product names. The useful unit is the decision support around each name: use case, constraints, pricing context where verified, implementation considerations, and sources that let the reader check a claim. If those elements are absent, changing the first-ranked product does not fix the underlying credibility gap.

This does not mean a company cannot publish a comparison involving its own product. It means the page must earn attention beyond its self-interest. A vendor can explain where it fits, who should not choose it, and what objective criteria shaped the comparison. Those admissions often make a commercial page more useful because they give the reader a way to test its conclusion.

Local and service businesses should take the same lesson without copying SaaS formats. A page declaring itself the best local provider is weak evidence on its own. Service businesses need accurate service coverage, current location information, qualifications, and clear descriptions of what a customer can expect. Those details help both readers and systems assess relevance without asking the business to certify itself.

1. B2B SaaS vendors with self-ranked best-of pages are directly represented in the reported sample.
2. Agencies scaling templated comparison pages should examine whether each page has independent decision support.
3. Publishers can compare their own offering, but should disclose their role and show the method.
4. Local businesses should prioritize accurate service and location evidence over unsupported superlatives.

Does Google prohibit self-ranking listicles?

Google does not state that self-ranking listicles are prohibited, and the reported finding should not be read as a new published ban. Google's documentation says there are no special requirements to appear in AI Overviews or AI Mode beyond being indexed and eligible for a Search snippet. It also says that following foundational SEO practices and creating helpful, reliable, people-first content remains the recommended approach. Google Search Central, 2025-12-10

The more relevant question is whether the page gives the system and the reader reasons to trust its conclusions. Google states that AI Overviews and AI Mode surface relevant links, and that these has can show different links because they may use different models and techniques. No page has a guaranteed place in a response merely because it satisfies technical requirements. Google Search Central, 2025-12-10

Ray's result is therefore better treated as an observed performance change, not a rule to reverse-engineer. Chasing an alternate phrase, a markup pattern, or a new page template would miss the point. A publisher that wants durable eligibility should improve the evidence on the page, identify who reviewed it, and maintain it when product facts change.

Google also says special AI text files or special schema.org markup are not required for these features. Structured data can still be appropriate when it accurately describes content, but it cannot turn an unsupported self-ranking claim into independent evidence. The work is editorial before it is technical.

1. No Google documentation cited here bans self-ranking listicles outright.
2. Google says there are no additional technical requirements for AI Overviews or AI Mode.
3. Eligibility does not guarantee that Google will serve a page or cite it.
4. Accurate structure helps discovery, but it cannot substitute for substantiated content.

How should publishers rebuild a comparison page?

Publishers should rebuild a comparison page around a transparent decision method rather than a predetermined winner. Start by stating the audience, the job being evaluated, and the criteria used to compare options. A buyer looking for enterprise governance is solving a different problem from a small team that needs a simpler workflow. A useful page makes that difference visible before it introduces any brand.

Next, separate facts from editorial judgment. Product capabilities, pricing terms, integration details, and policy statements should link to primary documentation when available. Editorial assessments should explain what was tested or reviewed, the date of the review, and the limits of that process. When the publisher sells one of the products, disclose that relationship plainly and avoid treating its own product copy as independent proof.

Then give each option a meaningful profile. Do not create one generic paragraph with a product name swapped in. Explain the use case it serves, its constraints, and the type of buyer for whom it is a poor fit. This approach is harder to scale, which is precisely why it is more useful. It gives an AI answer distinct details to cite instead of a repeated claim to ignore.

Finally, maintain the page. Comparison content degrades quickly when plans, features, ownership, or documentation change. Set a review date, keep a record of source checks, and remove claims that cannot be reverified. Freshness is not cosmetic when the page asks readers to make a decision.

1. Define the audience and decision criteria before ranking products.
2. Link factual claims to primary sources and distinguish them from editorial judgment.
3. Disclose the publisher's commercial relationship to any product discussed.
4. Review the page on a schedule and remove claims that cannot be verified.

What should content teams measure after this change?

Content teams should measure source appearance and recommendation separately after this change. The key question is not only whether an AI Overview links to a page. It is whether the answer cites the brand, names it favorably, or recommends a competitor after drawing on the page. Without that separation, a team can celebrate a citation while missing the commercial outcome.

Build a stable prompt set for the category and keep the prompts, locations, dates, and surface consistent. Record the full answer, visible links, named brands, and the landing pages cited. Repeating the same set over time turns anecdote into a trend line. It also makes it possible to identify whether an editorial revision changed citation behavior or whether the result moved across the broader category.

Use clear terms for each metric. Citation Count per day is volume. Answer Presence is breadth across the question universe. Share of Voice is relative visibility against named competitors. Citation Share is the headline percentage of relevant AI answers that cite the brand. Mixing these measures creates false conclusions, especially when one page can be cited while a competitor receives the recommendation.

Do not use a single volatile answer as proof that a strategy works or fails. Google says AI Overviews and AI Mode can vary because they use different models and techniques. A useful measurement program looks for repeated patterns across a defined prompt set, then reviews the pages behind those outcomes. Google Search Central, 2025-12-10

1. Track citations, named recommendations, and competitor recommendations as separate outcomes.
2. Keep prompts and collection conditions consistent over time.
3. Use Citation Share for the percentage of relevant answers that cite the brand.
4. Investigate repeated patterns rather than reacting to one answer.

What is the practical response for a self-ranking listicle already live?

The practical response is to audit the page for unsupported self-promotion, then decide whether it can become a credible comparison. Begin with the headline, opening claim, product order, and call to action. If the page declares a winner before showing criteria, treat that as a finding. If every option sounds similar except for the publisher's product, treat that as a finding too.

Keep pages that have a real editorial method and enough evidence to improve. Add a clear disclosure, describe the selection criteria, cite primary sources, and make limitations visible. Reorder products only if the evidence supports the new order. The goal is not to hide commercial intent. The goal is to give a buyer useful information even when the publisher's product is not the right choice.

Consolidate or retire thin variants. Multiple city, category, or competitor pages that repeat the same unsupported claim can create more maintenance burden than useful coverage. A smaller set of pages with distinct research and current sources is easier to defend, update, and measure.

The before-and-after evidence from Ray's work should prompt a review, not panic. Her reported decline is a signal from a limited sample, and Google has not published a rule saying vendor listicles cannot appear in AI Overviews. The reliable response is better editorial evidence, clearer provenance, and disciplined tracking of whether the answer actually cites or recommends the brand. PPC Land, 2026-10-05

1. Audit the page's method, evidence, disclosure, and treatment of competitors.
2. Improve pages that can give readers a defensible comparison.
3. Consolidate or retire thin pages that only restate the publisher's preferred conclusion.
4. Track outcomes after revisions before deciding that the change succeeded.

Sources

Source: PPC Land

Lily Ray's September 2026 research reported 38% fewer AI Overview citations of vendor-written self-ranking pages than June, and reported an 83% omission rate for the self-ranking brand across 100 B2B software queries. PPC Land, 2026-10-05

Source: Google Search Central

Google says there are no additional requirements or special optimization needed to appear in AI Overviews or AI Mode, and that normal Search technical requirements and SEO best practices apply. Google Search Central, 2025-12-10

Source: Google Search Central

Google explains that AI Overviews and AI Mode may use query fan-out and can show varying links because they use different models and techniques. Google Search Central, 2025-12-10