For the past couple of years, one of the most reliable ways to influence AI search answers has been the self-promotional listicle: a “best X software” article published by a brand that, unsurprisingly, ranks itself first. It worked because it exploited a content gap — most companies were too self-conscious to publish something that openly biased in front of human readers, so almost nobody was competing for those queries. AI systems, with no sophisticated way to separate genuine authority from self-endorsement, cited the pages and often repeated the recommendation.

New research tracking 100 B2B “best” queries in Google’s AI Overviews across three checkpoints between April and June 2026 suggests that window is closing. Google appears to have changed how it weighs these pages, and for most brands, publishing one is now a liability rather than an advantage.

Getting cited is not the same as getting recommended

AI search analysis tends to collapse into a single metric: citations. But a citation only means a page was used as a source to build the answer. A recommendation means the model actually named your brand as a pick. Those are different outcomes with very different business value, and the research shows they are increasingly decoupled.

Across the tracked queries, a brand’s own “best” listicle was cited as a source in 74% of the prompts that returned an AI Overview. But when that citation happened, the self-promoting brand was left out of the actual recommendation 69% of the time — 224 of 323 citations. In other words, publishing the article got the page pulled into the answer, but the brand naming itself the winner rarely got the AI system to agree.

This matters more than it sounds. Pew Research has found that when a Google search surfaces an AI summary, users click through to a source link only about 1% of the time. Citation without recommendation delivers almost none of the traffic upside people assume it does — you’re contributing to an answer that names your competitors instead of you.

The mechanism working against self-promoters

The pattern in the data is consistent: a self-promotional listicle that mentions competitors by name — even briefly, to “compare” against them — functions as a vote for those competitors. Several documented examples make the mechanism concrete. In one case, a learning management system ranked itself the best LMS for selling courses, was cited heavily in the AI Overview, and was still excluded from the recommendation, while the competitors it named in its own article — Kajabi, Thinkific, LearnWorlds, and Teachable — all made the cut. The same pattern repeated for help desk software, task management tools, and survey platforms: the self-promoting brand supplied the citation, and its own mentions of established rivals fed those rivals directly into the recommendation. You will find more ideas in is Google finally cracking down on self-promotional.

Why do some brands still get recommended when they self-promote? Authority, not page quality. Comparing referring domains and AI mention volume between recommended and excluded brands in the same category shows a consistent gap: category leaders with far more backlinks and far more third-party AI mentions get the benefit of the doubt, while lower-authority brands publishing the identical tactic get cited and passed over. The listicle itself is often well-built and well-optimized — that’s precisely why it ranks and gets pulled in as a source — but page quality doesn’t compensate for a thin off-site reputation.

The real cost: Google is treating the tactic as a risk signal, not just ignoring it

This isn’t only a recommendation problem. Around January 20, 2026, a cluster of sites — many of them heavy users of self-promotional listicles alongside AI-generated content at scale and formulaic comparison pages — saw a sharp, sustained drop in organic visibility, and the damage was not confined to the offending subfolder. Multiple sites saw the decline spread across the entire domain, with drops accelerating through the May 2026 core update. That is the pattern you’d expect from a broader spam signal, not a narrow content penalty: the tactic doesn’t just fail to help, it appears to actively damage domain-wide trust once it’s used at scale.

Google’s AI Overviews are also starting to editorialize around the problem. Certain “best expert” queries in crowded categories now return answers that flag the space as “saturated with self-proclaimed experts,” pointing users toward better evaluation criteria instead of a ranked list. Other assistants show similar caution when asked about categories known for this kind of manipulation. That is a direct signal that language models are getting better at recognizing when a category has been gamed — and are starting to warn users about it rather than simply repeating the loudest self-endorsement.

Who actually gets cited for “best” queries

A handful of high-authority, review-driven, user-generated sources dominate citations for queries containing the word “best”: Forbes, Reddit, and YouTube consistently outrank individual brand listicles, and Reddit’s share has grown noticeably in recent months. That tells you where AI systems default to when they don’t trust brand self-reporting — independent, aggregated opinion. If you’re trying to influence a “best” answer, competing for space on those platforms is a better use of effort than another self-published ranking article.

What to do instead of publishing another self-ranking listicle

The lesson isn’t “stop writing comparison content.” It’s that self-endorsement doesn’t substitute for third-party corroboration, and in enough cases now actively works against you.

• Stop naming competitors in content that ranks you first. If you must publish a comparison page, do it without handing AI systems a ready-made list of category leaders to recommend instead of you.
• Invest in being talked about elsewhere. Reviews, comparison roundups on independent sites, forum threads, and YouTube coverage build the referring-domain and mention volume that actually separates recommended brands from merely cited ones. Our guide on why AI recommends your competitor and what to do about it covers this in more depth.
• Audit what your existing content is inadvertently promoting. If your comparison or “alternatives” pages mention rivals by name, check whether they’re functioning as free advertising in AI answers. We break this down in is your content strategy accidentally recommending your competitors.
• Track recommendation, not just citation. If your AI visibility reporting stops at “were we mentioned,” you’re missing the outcome that matters. See what it takes to get cited, and stay cited, in AI search for a fuller measurement approach.
• Build genuine category authority. Backlinks, third-party mentions, and consistent brand presence across the sources models already trust move the needle more than any single page. Our GEO strategies to get AI search engines to recommend your brand lays out where to focus that investment.

If you’re an established brand, the calculation is different

None of this means self-promotional listicles never work. Authoritative, well-recommended brands can still publish one and get both cited and recommended, because the model already has ample third-party evidence that they belong at the top of the category. But even then, the format carries reputational risk with human readers who land on an obviously self-serving page, and it’s increasingly scrutinized in industry coverage. For an established brand, the honest question is whether the marginal benefit is worth the trust cost — and given that other sites are likely already recommending you without prompting, the listicle may be adding less than it seems.

Frequently asked questions

Do self-promotional listicles still work in AI search?

Rarely, and mostly only for brands that are already category leaders with strong third-party authority. For most brands, the research shows the self-promoting listicle gets cited but the brand is excluded from the actual recommendation roughly two-thirds of the time.

Why would naming a competitor in my own content hurt me?

AI systems appear to treat a mention of a competitor within your content as a legitimate data point about that competitor, feeding it into the recommendation even when your own article ranks your brand first. You end up supplying evidence for rivals rather than for yourself.

Is citation frequency a reliable AI search metric?

Not on its own. Citation only measures whether a page was pulled into an answer as a source. Recommendation measures whether the brand was actually named as the pick, and the two increasingly diverge, especially for lower-authority sites.

Did Google penalize sites for using self-promotional listicles?

Multiple sites that heavily used this tactic alongside other scaled SEO/GEO tactics saw sharp organic traffic declines starting around January 2026, with the impact spreading beyond the specific pages to the wider domain and worsening through the May 2026 core update.