“Best CRM software.” “Top help desk platform.” “The #1 tool for X.” For a couple of years, one of the cheapest ways to show up in AI answers was to publish your own “best [category]” listicle and quietly rank your own brand at the top. It worked. Language models pulled these pages in as sources, and the brands behind them got cited. New analysis from Lily Ray, VP of SEO and AI Search at Amsive, suggests that trick has started to turn on the brands using it. In her framing, ranking yourself #1 in your own listicle can now act as a vote for your competitors.
The tactic, and why it worked
The industry name for this is the “self-promotional listicle.” You write an article titled something like “The 7 Best Project Management Tools,” and you put your own product at number one, ahead of the genuine market leaders. Before AI search, almost nobody did this, because openly biased content erodes trust with any human who reads it. The arrival of AI answers created a gap. Suddenly there was a content void around questions like “what is the best tool for X,” and self-ranking pages rushed to fill it.
For a while, that paid off, because early language models did not have a reliable way to separate self-promotion from genuine authority. If your page said you were the best and used the right format, you could get pulled in as a source.
How common did this get
Very. Ray reports finding 184 of these self-promoting listicle pages across 146 different brands, and notes that the format went viral in 2025 alongside the wider push into what the industry calls GEO. Even Shopify, she points out, at one stage had more than 100 articles of this type, and now appears to be culling many of them. When a tactic scales like that, the incentive for search engines to react scales with it.
What Lily Ray measured
Ray tracked 100 B2B “best [category] software” queries in Google’s AI Overviews and pulled the answers and their exact sources on three dates between April and June 2026. About 1 in 5 of those queries did not return an AI Overview at all. On the 80 that did, she separated two things: whether a brand’s own listicle was cited as a source, and whether that brand was actually recommended as a pick in the answer.
The gap between the two is the whole story. When a brand’s own self-promotional listicle got cited, that brand was still left out of the actual recommendation 69% of the time. That is 224 of the 323 self-promotional listicles she counted. Across the full set, 74 of the 100 tracked queries produced an answer that cited a self-promoter’s page but recommended someone else. The page gets used. The brand does not get picked.
The pattern is not a one-off
Ray shows the same shape across very different categories. A learning platform ranks itself best “LMS for selling courses,” gets cited, and the answer recommends Kajabi, Thinkific, LearnWorlds and Teachable instead. A help desk tool ranks itself best, gets cited, and the answer recommends Zendesk, Freshdesk and Help Scout. A time-tracking product ranks itself best “task management software,” gets cited, and the answer names Todoist, Asana, Trello and ClickUp. A survey tool ranks itself best across two of its own listicles, gets cited from both, and the answer recommends Qualtrics, SurveyMonkey and Typeform. Different industries, identical outcome.
The survey category is a good illustration of how little a citation buys you. A brand can be pulled in from two of its own self-ranking pages at once, so it shows up twice in the sources, and still not appear once in the list of tools the answer actually suggests.
A citation is not a recommendation
This distinction matters more than it looks, because a citation almost nobody sees is worth very little. Ray argues the recommendation matters “by an order of magnitude,” and the click behaviour backs the spirit of that. Pew Research found in July 2025 that when a Google search showed an AI summary, users clicked a link inside that summary in just 1% of visits. If almost nobody clicks the sources, then the words in the answer, the brands the AI names as good options, are the whole game. A self-serving listicle can earn you a citation nobody clicks while handing the visible recommendation to the rivals you listed underneath yourself.
It is worth being precise about what “recommendation” means here, because it is the currency that is quietly replacing the blue link. When someone asks an assistant “what is the best help desk for a small team,” they usually act on the two or three names in the answer. They do not scroll a page of ten options and form their own view. So being one of those named picks is not a nice-to-have on top of ranking. In an AI answer, it is the ranking.
Why the big brands still get away with it
Here is the uncomfortable part for smaller companies. Ray’s read is that the outcome depends almost entirely on how authoritative your brand already is. A category leader can publish “we’re the best” and still get both cited and recommended, because the rest of the web already agrees. A smaller brand doing the identical thing gets cited and skipped.
Her authority comparisons make the gap concrete. Looking at referring domains, and at how often each brand is mentioned in AI Overviews and in ChatGPT, the recommended brands sit far ahead of the cited-but-ignored ones on every signal. In HR software, the recommended names lead the cited-but-unrecommended ones across the board.
The CRM category tells the same story. Ray notes that the one self-ranking CRM brand that also gets recommended is Monday, sitting on a Domain Rating in the 90s and tens of thousands of referring domains, while challengers that publish the same kind of self-promotional page trail on every metric, some with zero AI Overview or ChatGPT mentions at all.
The pattern is consistent. The model recommends the brands the wider web already recommends, not the ones that shout loudest about themselves. Which means the lever that actually moves the answer is not your own page. It is everyone else’s.
It can also cost you in classic search
The AI answer is not the only exposure. Ray also reports that around 20 January 2026, dozens of sites leaning hard on this tactic began losing organic traffic, and that the declines continued through Google’s May 2026 core update, in many cases spreading across the whole domain rather than staying in one folder. The sites she analysed, more than 40 of them, generally combined several spam-adjacent signals at once: scaled AI-generated content, mass-produced format pages, and hundreds or even thousands of articles ranking their own brand #1.
Two honest notes here. This is Ray’s analysis of a set of sites she chose, not a Google announcement, and Google has not confirmed a mechanism. And the risk she describes is about scale and stacking risky tactics, not one comparison page. Treat it as a strong directional signal: pushing self-promotion at industrial volume can put your entire domain in the blast radius of a core update.
Google is starting to warn searchers directly
There is a third consequence that should give any “we’re the best” strategy pause. Ray shows that for some “best [x]” queries, Google’s AI Overview now attaches a disclaimer, telling searchers that a category is “saturated with self-proclaimed experts” and steering them toward professionals recognised for measurable results. She notes that Claude does something similar, flagging that certain categories have been spammed. When the search engine and the assistant are both coaching users to distrust the word “best,” building your visibility on that exact word looks less clever by the month.
So what does the AI actually trust
If not your own listicle, then whose? Ray’s citation data points at the usual high-authority names. For “best” queries in AI Overviews, Forbes, Reddit and YouTube keep climbing as the sources the answers lean on, with Reddit in particular growing fast and Forbes Advisor surging since around March. In other words, AI answers increasingly borrow their judgement from independent review sites and from places where real people compare options in public.
What Google itself says about this
None of this contradicts Google’s own advice, which is the part worth sitting with. In its guide to optimising for AI features in Search, Google is blunt that chasing “mentions” across the web in an artificial way is not the shortcut it looks like, and that manufactured or inauthentic placements are the kind of thing its spam systems are built to catch. Its standing message is the same one it has repeated for years: make content for people first, and earn a reputation genuinely rather than assembling signals to game a system.
Read against Ray’s data, that guidance stops sounding like boilerplate. A wall of pages calling yourself the best is, almost by definition, an artificial mention of yourself at scale. It is exactly the pattern Google says it discounts. The self-ranking listicle worked for a while because the models had not caught up, not because Google ever endorsed it.
How to tell if this is happening to you
You do not need a research budget to check your own exposure. A short audit gets you most of the way:
- Run your own “best [category]” queries. Ask Google and an assistant like ChatGPT for the best tools or providers in your space. Note which brands get named as picks. If you are not one of them, you are not being recommended, whatever your own pages say.
- Look at the sources under the answer. If your page appears in the citations but your brand is missing from the recommendation, you are living the exact split in this analysis. A citation there is not a win.
- Count your self-ranking pages. One honest comparison article is fine. If you have dozens or hundreds of pages that all crown your own brand #1, you are carrying the risk profile that Ray associates with domain-wide declines.
- Compare your off-site footprint to the brands that do get recommended. How many independent sites link to you or mention you, versus them? That gap, not your copywriting, is what the AI is reading.
If the answer is “cited but not recommended,” the fix is not to write the listicle more aggressively. It is to close the authority gap.
What earns the recommendation instead
The thing that actually moves the recommendation is boring and hard to fake: being talked about and linked to in places you do not control. In practice that means a handful of unglamorous, durable moves.
- Earn genuine third-party mentions. Get written about, reviewed, quoted and linked by sites that are not you: trade publications, independent reviewers, partners, journalists. This is the signal the authority tables above are really measuring.
- Get into the roundups you did not write. Being included in someone else’s honest “best of” list carries the weight your own list never will. Pitch reviewers, offer trials, make it easy to evaluate you fairly.
- Show up where buyers actually compare, honestly. Communities like Reddit, and platforms like YouTube, are increasingly cited by AI answers. You cannot spam your way in, but you can participate usefully, answer questions, and let real users vouch for you.
- Collect real reviews. Genuine ratings and testimonials on independent platforms feed the same reputation the model reads. Ask happy customers, make it painless, never fabricate.
- Build topical authority, not self-praise. Publish first-hand, specific, opinionated content that answers real questions in your field. Google itself says this is the kind of thing AI systems are built to surface, and generic self-ranking listicles are not.
- Tidy your entity signals. Make sure your brand is described consistently across your site, your profiles and structured data, so machines can connect the mentions you do earn to one clear entity.
You can keep a “best [category]” page if it is honest and genuinely helpful to a reader. Just stop expecting a self-crowned #1 to carry you. That earned-reputation work is the substance behind honest AI visibility, and it is the opposite of a self-ranking listicle.
The picture for local and multilingual businesses
Ray’s sample is US B2B software, but the mechanism travels, and for the smaller and local businesses we work with across Poland, Ukraine and the wider EU it travels in an unhelpful direction. If you are a regional agency, shop or service provider, you were never going to out-authority a global brand by declaring yourself the best in a blog post. Your realistic edge is local and specific: reviews from customers in your city, mentions in local and trade media, listings and profiles that agree with each other, and content that answers the precise questions your buyers ask in their own language.
That last point matters more in multilingual markets. Authority does not transfer cleanly across languages, so a brand that is well recommended in English can be nearly invisible in Polish or Ukrainian AI answers, and a focused local player can win the local-language recommendation by being the option real people and local sites actually talk about. The work is the same as above, earned mentions and genuine reviews, just aimed at the sources that carry weight in your market rather than at a global leaderboard you cannot top.
Common questions
Should I delete my “best” listicle? Not necessarily. One honest, genuinely useful comparison page is fine, and deleting a single article will not transform your visibility. The risk Ray describes is about scale: dozens or hundreds of near-identical pages that all rank you first. If that describes your site, consolidating and pruning is sensible. If it is one page, focus your energy on earning authority instead.
Do these listicles still work for big brands? Ray’s data suggests they can, because an established brand already has the reputation to be recommended anyway. The listicle is not what earns them the pick, their off-site authority is. For a smaller brand without that footprint, copying the tactic mostly produces citations that go nowhere.
Is this only Google AI Overviews, or ChatGPT too? The measured data here is Google’s AI Overviews. But the underlying logic, that assistants lean on how much the wider web vouches for you, applies broadly, and Ray notes assistants like Claude adding their own “spammed category” warnings. Do not assume one channel is safe because another is being measured.
How long does it take to build real authority? Longer than writing a listicle, which is exactly the point. Earned mentions, reviews and coverage accumulate over months, not days, and there is no honest shortcut. Anyone promising instant AI-recommendation status is selling the next version of the tactic that just backfired.
Is any of this confirmed by Google? No, and we are careful to say so. This is one researcher’s analysis of a specific dataset. Google has not published a mechanism. What makes it worth acting on is that the direction matches Google’s own stated guidance and what practitioners see in the field, not that Google has ratified the numbers.
The honest caveat, and the takeaway
One researcher’s dataset is not a ruling from Google. Ray calls this her interpretation, and so do we: the sample is B2B software, it is US-centric, and Google has not published the mechanism. But the direction lines up with what Google keeps saying out loud, that it wants to reward genuine reputation over self-promotion, and with what we see in our own AI-visibility work.
If you have been leaning on self-promotional listicles, the fix is not another listicle. It is the slower, real work of becoming a brand the rest of the web actually recommends. Strong AI search visibility, like good SEO, has always come from earned authority rather than clever self-labelling. And if your growth plan quietly depends on scaled, self-serving content, it is worth reading how Google is getting better at spotting synthetic filler before the next core update reads it for you.
Sources
- Lily Ray, “Why Calling Yourself the ‘Best’ Could Be Helping Your Competitors Win in AI Search,” Amsive / Substack, 17 June 2026: lilyraynyc.substack.com
- Search Engine Land, “Google AI Overviews cite self-serving listicles, but recommend competitors 69% of the time”: searchengineland.com
- Google Search Central, “Optimizing for generative AI features on Google Search”: developers.google.com
- Pew Research Center, “Google users are less likely to click on links when an AI summary appears in the results,” July 2025: pewresearch.org
