You Wrote the Article AI Used to Recommend Your Competitor

Here's a scenario that's playing out right now across thousands of businesses. You spent months building out your content strategy. Blog posts, comparison pages, "best of" lists. All properly optimised. All ranking well. Google's AI Overview reads your article, pulls the structure and criteria you laid out, and then uses that framework to recommend three of your competitors instead.

This isn't a hypothetical. Lily Ray's analysis of 100 B2B queries across Google's AI Overviews found that 69% of the time a brand's own content was cited, that brand was NOT the one recommended. Your listicle becomes the scaffolding. Someone else gets the spotlight.

The old SEO playbook said: create comprehensive content, rank for it, capture the click. But when 68% of Google searches now end without a click, the click isn't the game anymore. The recommendation is. And AI systems have a very different set of criteria for who gets recommended versus who gets cited.

The Citation Paradox: Getting Credit Without Getting Chosen

Lily Ray examined what happens when businesses write self-promotional "best [category]" listicles and those pages appear in Google's AI Overviews. Across 80 prompts that triggered AI responses, self-promotional listicles were cited 323 times. But in 224 of those cases, the brand behind the listicle wasn't the one Google recommended.

The pattern is consistent. For a query like "best LMS for selling courses," Google's AI Overview cited Oasis LMS's comparison page. It used that page's structure, evaluation criteria, and category framing. Then it recommended Kajabi, Thinkific, LearnWorlds, and Teachable.

Why? Because the AI Overview didn't just read one page. It cross-referenced the criteria in that listicle against the broader web to determine which brands had the strongest signals across multiple independent sources. The brands that already led their categories, were widely mentioned by third-party sources, and had stronger presence across the web were the ones that surfaced in recommendations.

Your content becomes the map. AI uses it to find the destination. And the destination is whoever the rest of the internet already agrees is worth recommending.

This is the uncomfortable truth behind the content-first SEO strategy that dominated the last decade. Creating content about your industry isn't the same as being the business your industry talks about. The distinction between those two things has never mattered more.

What AI Actually Recommends (and Why Mental Availability Explains It)

Byron Sharp's research at the Ehrenberg-Bass Institute, across 130+ brands in 13+ product categories, established that brands grow by being easy to think of in buying situations. He calls this mental availability: the probability that your brand comes to mind when someone enters the category.

Sharp distinguished this from simple awareness. Awareness is binary: have you heard of us? Mental availability is multidimensional: do you come to mind across the full range of situations that trigger a purchase?

Here's why that matters now more than ever. AI systems are, in effect, operationalising mental availability at scale.

When ChatGPT, Perplexity, or Google's AI Overview generates an answer, it's doing something remarkably similar to what a human brain does under Sharp's framework. It's asking: across all the information I have access to, which brands are most frequently associated with this category, across the widest range of contexts?

The data bears this out. Brands that already led their categories, were widely mentioned by third-party sources, and had stronger presence across the web were more likely to appear in AI Overview recommendations. The AI isn't evaluating your claims about yourself. It's evaluating the consensus of the internet about you.

Jenni Romaniuk, Sharp's colleague at the Ehrenberg-Bass Institute, developed the concept of Category Entry Points (CEPs): the specific needs, occasions, and motivations that cause someone to think about a category. As Semrush's research on applying CEPs to AI search noted, AI prompts give us a direct window into these category entry moments. And the brands that get recommended are the ones with the broadest association across those moments, not the ones who wrote the most comprehensive guide.

This is Sharp's framework made measurable in a way it never was before. Your signal strength across the web, not the quality of your own content, determines whether AI recommends you.

Where AI Actually Looks (It's Not Your Blog)

If AI recommendations are driven by web-wide consensus rather than individual page quality, the obvious question becomes: where does that consensus form?

The data here is striking. Across 150,000+ citations analysed, the top sources cited by ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews are:

This maps directly to something Rory Sutherland has argued for decades. The perceived trustworthiness of information is shaped by who says it and where, not just what's said. A recommendation from a stranger on Reddit carries more weight than a brand's claim about itself, for the same reason a word-of-mouth referral from a neighbour carries more weight than a billboard. The source signals credibility in a way the content itself cannot.

For most small businesses, this is a blind spot. They've invested everything in their website and their own content. But their business exists in one place online while AI looks in a thousand. The businesses getting recommended are the ones leaving traces across Reddit threads, LinkedIn comments, YouTube videos, Google Reviews, industry forums, and directory listings.

The Five Traits That Actually Win in Search

If content volume isn't the answer, what is?

This aligns with what Rand Fishkin has argued about the death of the "ultimate guide". Broad, comprehensive content that anyone could write is increasingly worthless. What wins is content that comes from doing something nobody else has done: running the experiment, serving the customer, collecting the data, building the tool.

For an Australian tradie spending $3,000 a month on Google Ads, this doesn't mean they need to build a software tool. It means documenting real project results with photos and numbers. It means publishing their actual pricing methodology instead of a generic "factors that affect cost" blog post. It means creating a simple calculator or quiz that helps visitors understand what they need. These are all proprietary assets. They can't be replicated by a competitor writing from a template.

The pattern is clear: Google and AI systems are both moving toward rewarding businesses that own something unique over businesses that describe something generic.

Mentions Are the New Links

Ethan Smith, CEO of Graphite and adjunct professor at IE Business School, has spent the past year mapping what actually gets cited in ChatGPT, Claude, and Gemini. His conclusion: "The future of search isn't links. It's mentions."

This is a fundamental shift. For 25 years, SEO was built on the premise that links are votes of confidence. A link from a reputable site to yours passed authority. The entire industry of link building, guest posting, and digital PR was oriented around acquiring those links.

But AI systems don't follow links the same way Google's original PageRank algorithm did. They process text. When an AI reads a Reddit thread where someone says "We used [Business Name] for our office fit-out and they were excellent," that mention carries weight regardless of whether it's hyperlinked. The AI is reading the sentiment and frequency of name-drops across the entire web, not counting backlinks.

This is why the 68% zero-click figure matters so much. If most searches don't result in clicks, then the value of being mentioned in the answer (whether or not it drives a visit) is the primary outcome. Being named in an AI response is a brand impression. And when AI search traffic does result in a click, early data suggests it converts at 5-9x the rate of traditional organic search, precisely because the AI has already done the evaluation and narrowed the field.

The implication for SMEs is significant. You don't need to be everywhere. But you do need to be mentioned in the right places. A single genuine recommendation in a Reddit thread with 50+ upvotes has more AI citation value than ten blog posts on your own website. A LinkedIn comment where you share a specific, non-obvious insight from your work has more value than a press release.

This connects to a broader point we've explored before about how Google now ranks reputation rather than pages. The shift from links to mentions is the mechanism behind that reputation-first approach.

The Strategy Most Businesses Get Backwards

Here's what typically happens. A business hires a content writer or an agency. They produce blog posts targeting keywords. The posts are well-written, properly structured, maybe even based on real expertise. Traffic comes in. Some of it converts.

Then AI Overviews roll out. The content still ranks. It even gets cited in AI answers. But the business notices something odd: their competitor, who blogs far less, keeps showing up in the AI's actual recommendations. The competitor gets the "we recommend" line. They get the footnote.

The difference? The competitor has 200+ Google Reviews. Their founder is active on LinkedIn sharing specific project stories. A customer mentioned them in a Reddit thread three months ago. They spoke at a local industry event that was written up in a trade publication. They have a simple cost calculator on their website.

None of that is "content marketing" in the traditional sense. But all of it is exactly what AI systems weigh when deciding who to recommend.

The old formula was: create content > rank for keywords > capture clicks.

The new formula is: build something worth talking about > get mentioned across the web > become the business AI recommends.

The first formula is a content strategy. The second is a brand strategy, which is exactly what marketing science has always said drives growth. Byron Sharp's work showed that penetration, not loyalty, drives brand growth. The AI recommendation engine is an accelerant for that same dynamic: the brands with the broadest presence win more recommendations, which drives more awareness, which creates more mentions, which drives more recommendations.

What This Means for Your Business

Stop writing articles that position your competitors. If you've published "Best [Category] in [City]" listicles that name your competitors, you've built a research brief for AI to evaluate them against you. Unless you're already the dominant brand in that category, you'll lose that evaluation.

The businesses winning this shift aren't the ones producing the most content. They're the ones that other people's content mentions. That's always been the definition of a strong brand. The difference is that AI systems have now made it the definition of search visibility too.

Further Reading

• 5 Data-Backed Features of Websites Winning Google in 2026 - Cyrus Shepard's analysis of 400+ sites identifying the traits that predict organic traffic outcomes
• Google AI Overviews Cite Self-Serving Listicles, But Recommend Competitors 69% of the Time - Lily Ray's study on the citation-recommendation gap in AI Overviews
• When Google Stops Sending Clicks, What Still Works? - Rand Fishkin on the 68% zero-click reality and brand-building as the response
• Why Category Entry Points Belong in Every AI Search Strategy - Applying Byron Sharp's mental availability framework to AI search
• AI Citation Source Index 2026 - Data on the top 50 most-cited sources across major AI platforms