Self-promotional listicles can get your page cited without getting your business recommended. That's the distinction behind Lily Ray's June 2026 research on Google AI Overviews, and it changes how a “best of” article should be judged. If an answer uses your list but names your competitors, the citation alone doesn't establish a commercial win. The evidence is less settled on whether self-ranking causes a penalty. Here's what the research supports, what cuts against a blanket warning and why Parabolic Studio is adopting a more transparent way to publish lists.
What Is a Self-Promotional Listicle?
A self-promotional listicle is a comparison article in which the publisher sells in the category it covers and presents its own company or product as a leading option, often ranking itself first.
The conflict is straightforward: the business making the recommendation benefits if the reader accepts it. A clear disclosure helps the reader assess that conflict. Published criteria and useful comparisons help further. Neither turns the page into an independent review, but they give a buyer more to work with than a first place badge awarded by the company wearing it.
Why Brands Started Ranking Themselves First
These pages match a valuable buying question. Someone asking for the best software, agency or provider is already comparing options, and a well organized list can help them do it. For the publisher, inclusion brings its brand into a relevant comparison. The temptation is to go one step further and make the publisher the automatic winner, regardless of the buyer's needs.
That shortcut confuses two different jobs: explaining a category and proving superiority within it. A company may be well qualified to describe the options without having tested every competitor. A useful article makes the limit visible. It explains where its information came from, identifies the buyers each option might suit and avoids implying that an editorial order is the result of an independent performance test.
What Changed in 2026
Visibility declines were observed, but the cause is contested
Ray reported visibility declines from around 20 January 2026 among sites publishing many self-ranked lists. Those sites often had other potential problems, including scaled content and large sets of similar pages. Google hasn't confirmed an update specifically targeting self-promotional listicles. The observation warrants scrutiny, but it doesn't isolate self-ranking as the cause of the decline.
Cited, but not recommended
Her separate AI Overview study tracked 100 software queries at three checkpoints: 15 April, 15 May and 8 June 2026. Among the cited self-promotional lists, the publisher's brand was absent from the recommendations in 69% of cases. That result concerns a defined sample on Google's AI surface, not every category or answer engine.
The practical lesson is about measurement. A citation shows that your page was used as a source. A recommendation shows that your brand was offered as an option. Those outcomes can diverge. If your goal is to enter a buyer's shortlist, count the recommendation separately and examine the answer itself before calling a source link a success.
“Citations and recommendations can behave quite differently”.
Lily Ray, 17 June 2026
Google's rules now cover AI answers
Google's spam policies explicitly cover attempts to manipulate generative AI responses in Search. That is a broader rule than “don't write lists.” Disclosure or alphabetical order doesn't automatically establish compliance. The content still needs to be accurate and useful, and a network of thin pages created mainly to manufacture recommendations deserves a different assessment from a researched comparison.
| Cited | Recommended | |
|---|---|---|
| What it means | Your page is listed as a source the answer drew on | Your brand is named in the answer as an option to consider |
| What to assess | Whether the page provides useful evidence for the answer | Whether the brand is presented as a relevant choice |
| How to measure value | Referral visits and the context of the source link | Relevant mentions, enquiries and contribution to buyer consideration |
The Evidence Against Our Argument
There are good reasons to resist a simple claim that self-ranking no longer works. The evidence concerns different outcomes, comes from different samples and doesn't establish one rule across all answer engines.
- Many vendor lists still gain traffic. In his 24 August 2026 analysis, Kevin Indig examined 5.32 million search results. He reported that 62% of vendor authored listicles gained estimated organic traffic since January. That challenges the idea of a blanket demotion.
- Traffic and recommendations aren't interchangeable. Organic traffic growth doesn't show whether an AI answer recommends the publisher. Equally, being absent from a recommendation doesn't make a page worthless to readers.
- Established brands can still be recommended. Ray's findings include publishers whose brands were named. Her study doesn't establish that self-ranking prevents a recommendation or that every smaller brand is disadvantaged.
- Our own prompt record doesn't settle causation. Parabolic appears in a recorded “best AI SEO agencies in Canada” answer but not the corresponding “top” answer. That shows variation. It doesn't establish whether our Vancouver list helped, harmed or had no effect on either output.
These findings can coexist. Estimated search traffic, citations and brand recommendations measure different things. A page can gain traffic and still fail to get its publisher recommended for a particular question. It can also help some readers without proving that the publisher belongs first. That is why the decision should rest on the value of the comparison and the quality of its evidence, rather than a universal claim about algorithmic reward or punishment.
Our Position: Don't Rank Yourself First
Our editorial choice is to avoid ranking ourselves first in our own category lists. We want a reader to understand the options and assess our inclusion without having to discount a self-awarded victory. That is a judgment about credibility and usefulness. The research strengthens the case for measuring recommendations carefully, but it doesn't prove that moving our listing down will improve AI visibility.
A company can include itself where it's a relevant option. It should explain that it publishes the article, state what the comparison can establish and apply the same criteria to its own entry. Where the information supports fit rather than superiority, group the options by use case. Our generative engine optimization work follows the same broader aim: make a business understandable and support its claims with evidence readers can assess.
The Rules We're Adopting for Our Own Lists
- Say who published the list in the first three sentences, and say if we're on it.
- Never put ourselves first. Order lists alphabetically or group them by who each option suits, and say which.
- Show where the shortlist came from, for example the AI answer to a stated prompt on a stated date.
- Check every entry on its own website, and link there rather than to a directory profile.
- Give every entry, including ours, one honest limitation.
- Keep the year out of the title. Update the content and date the update, rather than refreshing the headline.
- Publish few lists, only in categories we actually work in, and never as programmatic variants.
- Re-check each list when the AI answer changes, and remove entries that no longer offer the service.
Our existing Vancouver agency list shows why this needs an honest transition. Checked on 3 October 2026, it still ranks Parabolic Studio first and includes 2026 in its title. It doesn't yet match the approach described here. We should revise that page before presenting these rules as a sitewide practice. Acknowledging the inconsistency is more useful than implying that the cleanup has already happened.
The companion Canadian agency shortlist is organized by business fit, with an alphabetical overview and a disclosure at the start. Parabolic is included without taking first place. That gives the reader a way to compare candidates while keeping our commercial interest visible.
What to Do If You Already Publish Self-Ranked Lists
If you already have self-ranked articles, review them by usefulness and outcome. A good comparison may be worth keeping. A page that repeats claims without evidence may need a substantial rewrite. Begin with the pages buyers actually see, especially those that attract enquiries or appear as sources in AI answers.
Check the disclosure, criteria and factual accuracy before changing the order. Then ask what the page promises. If it claims a ranking, does the evidence justify that ranking? If it offers a shortlist, would grouping by buyer need be clearer? Keep useful comparisons, merge thin variants and remove unsupported superlatives. Change a year in the headline only when the underlying content has been reviewed and materially updated. A recent looking title is no substitute for current information.
Measure the effect on your business separately from the effect on the page. ChatGPT brand mentions and earning ChatGPT recommendations explain the organic outcomes to look for. Use a consistent measurement framework or a quick visibility check to establish the baseline. For Google specifically, our AI Overviews guide covers the wider work. Reordering a list is an editorial improvement, not a complete visibility strategy.
Frequently Asked Questions
What is a self-promotional listicle?
A self-promotional listicle compares businesses or products in a category where the publisher also competes, usually presenting itself as a leading choice. Ranking the publisher first creates a visible conflict of interest. Disclosure, consistent criteria and evidence help readers assess the page, but don't make it independent.
Does Google penalize self-promotional listicles?
Google hasn't confirmed a penalty specifically for this page format. Researchers have observed declines at some sites using it heavily, while broader research shows many vendor lists gaining estimated traffic. Google's rules prohibit manipulative practices, including attempts to manipulate generative AI responses. The quality and purpose of the content matter.
Do self-promotional listicles still work in ChatGPT?
They can appear in sources, but a citation doesn't prove that they caused a recommendation. Research on Google AI Overviews shouldn't be treated as a universal ChatGPT rule. Check whether your business is actually named for relevant buyer questions, and repeat the checks rather than relying on one favorable answer.
Is it ever okay to include your own company in a "best of" list?
Yes, if your company is a relevant option and the commercial relationship is disclosed. Apply the same criteria to yourself, explain the evidence and give readers a meaningful limitation. An alphabetical overview or grouping by use case can make the comparison useful without implying an independent ranking.
Why would a listicle get cited but the brand not be recommended?
Using a page as a source and selecting a company as a recommendation are different outcomes. An answer can draw on a comparison while naming other options. Track the source link, the brand mention and the recommendation separately so you can see what the article is actually contributing.
A “best of” article should help someone make a better decision. If its main purpose is to put the publisher first, the comparison is doing too little work. Our approach is to disclose the conflict, explain the shortlist and organize options around buyer needs. That won't guarantee an AI recommendation, and the research doesn't justify promising one. It does give readers a clearer account of what they're looking at. Keep measuring the business outcome, and make the page useful even when an answer engine never cites it.
Parabolic Studio helps businesses improve how they're understood and represented in AI search. Our AI SEO and search visibility service connects that work with website improvements and reporting, so the goal stays focused on useful business visibility rather than a flattering screenshot.
