TL;DR: when a brand cites itself in its own "best software" listicle, Google AI Overviews does use the page as a source, but recommends a competitor instead 69% of the time. That's the finding from a Lily Ray analysis covered by Search Engine Land on June 18, 2026. The GEO lesson: being cited is not being recommended. Stop the self-serving listicles, build third-party authority.
On June 18, 2026, Search Engine Land covered an analysis by Lily Ray, a well-known SEO consultant, that exposes a GEO blind spot: when your brand publishes a "best [category] software" list with itself at the top, Google AI Overviews happily cites the page as a source, but recommends a competitor to the user in most cases.
The exact figure: across the queries studied, when a self-serving listicle was cited, the brand that published it was left out of the recommendation 69% of the time (224 of 323 citations).
What the study actually found
Lily Ray tracked 100 B2B "best [category] software" queries in Google AI Overviews across three measurement dates: April 15, May 15 and June 8, 2026. Data was pulled via Ahrefs Brand Radar. The sample covers 184 self-promotional listicles published by 146 brands.
Ray's conclusion is blunt: "Google appears to have decoupled what it cites from who it recommends." In other words, a page can serve as the factual source for the AI answer while the competitors listed inside that same page earn the recommendation. You write the article, your rivals pocket the lead.
Why this is a strong GEO signal
For two years, part of the industry has turned the "why we're the best" page into an AI-ranking weapon. The idea: cite yourself enough to get picked up by the models. This analysis shows the mechanism doesn't work as intended, and can backfire.
The core issue is the distinction between citation and recommendation. A citation is the AI pulling a data point from your page. A recommendation is the AI telling the user "pick this brand." The two are not the same. It's the same trap we flagged about the rise of zero-click search: showing up is no longer enough, you have to convert that appearance into intent.
Want to know whether AI cites you, or cites your competitors in your place? Cicéro runs a GEO diagnostic that measures your real presence in Google AI Overviews, ChatGPT and Perplexity answers, and flags the pages working against you.
What it changes for businesses
If your AI-visibility strategy rests on lists where you rank yourself #1, Lily Ray's analysis is a clear warning. For a smaller brand, the effect can be counterproductive: your page becomes a springboard for better-established competitors who are already cited elsewhere. Three moves to make now:
- Drop the disguised self-citation. A "best X" list that puts you on top with no transparent criteria won't win you the recommendation, and it erodes your editorial credibility.
- Build third-party authority. What carries weight is mentions on sources you don't control: independent comparisons, reviews, media citations. It's the same pattern we saw when Claude favors specific sources for its citations.
- Produce non-replicable content. Proprietary data, field insights, angles the AI can't fabricate on its own. That's the difference-maker as the AI search market fragments across ChatGPT, Gemini and Claude.
Our take
At Cicéro, we've said it since GEO began: optimizing to be cited is half the job. The other half is earning the recommendation, and that can't be gamed with a self-serving page. Lily Ray's data confirms it: Google can tell the difference between a source and a brand worth trusting.
Our advice fits in one line: write to be useful to the user, not to flatter your own ranking. It's the only version of AI visibility that holds up over time.
Sources
- → Search Engine Land: coverage of the analysis (June 18, 2026)
- → Lily Ray, newsletter: original analysis (100 B2B queries, measured April to June 2026)
Growth and SEO content strategist, I founded Cicéro to help businesses build lasting organic visibility, on Google and in AI-generated answers alike. Every piece of content we produce is designed to convert, not just to exist.
LinkedIn