- An Australian SEO educator says his new real estate site knocked a 10-year-old competitor off its own brand name ranking, using Claude Code for on-site analysis, with zero backlinks pointing at the page.
- Fan-out queries, the internal sub-questions AI search engines run before generating any answer, are becoming the blind spot that decides who gets cited and who disappears from AI-generated results entirely.
The founder of the AI Ranking community in Australia and owner of the YouTube channel @Nico_AIRanking released a video on July 24, 2025, that described how he outsmarted the competitor on their very own branded keyword. He posted the same description of the trick in the AI Ranking community post, where he outlined the basic principle: you do not compete with a competitor everywhere, but at their weakest point. The competing site, Propiedades Accent, had been live for more than a decade. His property site was only a few months old, had no backlinks to that specific page, and still climbed above them. When the method is spelled out precisely, brushing it off gets difficult.
Four Gaps, and Most SEOs Only Ever Patch One
Nico’s method works across four distinct weak points: on-site fundamentals, content depth, backlinks, and AI citations. His position is clear, sequence matters, you work through them in order rather than skipping ahead to backlinks, which is where most people land first. He ran Claude Code alongside Data for SEO to generate a scored head-to-head between his site and the competitor, covering five technical categories: title tags, heading structure, structured data, page speed, and image SEO.
His site won across every category. No H1 on the competitor’s page, no schema markup, no meta description, and images sitting with empty alt text. Any single one of those gaps in isolation would not have moved the needle. Piled together, they carved out enough space for a newer site to walk straight through. Fixing multiple mid-weight technical issues rather than chasing a single fix is not just his preference; it aligns with how Google’s core systems actually weigh pages against each other. As documented in a real Claude Code SEO case study, a traditional competitive analysis that takes two or more hours manually compresses to under 20 minutes when Claude Code handles the research and gap identification simultaneously.
With a 200,000-token context window, Claude allows practitioners to feed in a complete technical audit, a website architecture of a rival firm, and a branding guideline in one go. Work which would have taken three to four hours in the past is now completed within minutes. What’s more, Claude performs the work at a consistent level of quality regardless of who conducts it. As our coverage of AI search exposing weak SEO strategies notes, sites that rely on thin audits and irregular updates are finding those gaps harder to hide as AI-driven ranking signals become more demanding.
Content Is Not a Volume Problem, It Is a Structure Problem
The second pillar is where things get sharper. Beating Property Partners, a company operating across 50-plus offices worldwide with 402 backlinks pointing at their competing page, had nothing to do with publishing more. Nico’s page carried fewer property listings. What it carried instead was a region guide, a five-year pricing overview for the local real estate market, financing details for buyers, and FAQ content answering questions Property Partners had never bothered to address.
Datawise gave him both People Also Ask data and fan-out queries linked to his seed keyword. The gap between those two categories is not a small one. PAA shows what people type into search. Fan-out queries are what the AI engine asks itself internally, before it pulls anything to construct a response, breaking a single prompt into multiple sub-queries and running them simultaneously across the web. Going after that second category is what put his domain inside AI-generated answers while the competitor sat completely absent. Research from Surfer SEO, based on an analysis of 173,020 URLs, found that pages ranking for fan-out queries are 161% more likely to be cited in Google AI Overviews than pages targeting only the main query.
Research into how AI engines like Claude retrieve and cite content backs the reasoning. According to ClickRank’s analysis of Claude for SEO, these systems pull toward pages with case studies, original data, and answers detailed enough to signal real expertise. Thin content, regardless of keyword placement, gets passed over; not flagged as spam, but quietly deprioritized because the AI sees it as unreliable compared to a more current or more precise source. That is a different kind of filter, and most content strategies are not built to clear it.
Backlinks Still Count, Just Not for the Reasons Most People Cite
Nico’s list of four factors for judging backlinks is: Who the link site is; the relevance of the link page; the anchor text and context; and the spam level. The other guy had 17 referring domains. Every last one used brand name anchor text and each scored zero authority on Datawise. Six to seven strong contextual links, by his estimate, would be enough to take them.
He started with local citations, the free directory listings most businesses quietly skip past. Bing Places, Foursquare, Yelp, a Facebook Business listing, and Nextdoor. He maintained his efforts here until brand-name impressions began appearing in Search Console, then shifted to page-level linking to particular keywords. The order is not incidental. Chasing page-level links before your brand has any search presence usually burns the budget without moving rankings.
Nico was straightforward about outsourcing targeted link building when a situation calls for it, pointing to a $1,000 per month foundation package from Get Me Links for competitive scenarios. For agencies trying to put a number on this kind of work, that figure lands somewhere useful.
AI Citations Are the Weakness Almost Nobody Is Checking
Nico called AI citations the pillar most practitioners have not touched yet, and what he pulled from the data holds that up. His domain was showing up inside AI-generated answers across multiple queries. The competitor, carrying more backlinks and years of history, had zero citations, zero AI impressions, and zero AI search volume anywhere.
Fan-out queries explain the gap. When someone asks “Can I buy land in Chile?”, the AI search engine does not run a single lookup. It generates sub-questions internally, things like “Can a US citizen buy land in Chile?”, and draws from multiple sources before it writes a response. Nico had a blog post that answered that sub-question directly and in depth. The competitor did not. The citation landed on his page.
Research on how AI engines decide what to cite confirms the same pattern. Freshness, specific numbers, and answers structured to match the detail level of sub-queries all feed into that decision. According to ClickRank’s breakdown of AI citation behavior, a page running 2024 price data in 2026 gets flagged as potentially unreliable, regardless of how strong the domain is; outdated pages are often treated as a hallucination risk and bypassed in favor of more recent sources. Stale specifics get penalized even when the broader topic coverage holds up fine.
GrackerAI’s GEO Heist tool tracks which competitors are currently being cited by ChatGPT, Perplexity, Claude, Gemini, Copilot, and Google AI Overviews, and maps the content structure that earned each citation. Knowing exactly which queries your competitor shows up in, and what that cited page looks like structurally, is the intelligence Nico used to build his AI citation approach. As our earlier reporting on ChatGPT consuming 92% of AI referral traffic showed, the gap between being cited and being invisible in AI-generated answers is no longer a minor traffic variable.
What Shifts for Marketers Who Pay Attention?
None of what Nico describes is unfamiliar to experienced SEOs at the concept level. Where things have genuinely shifted is execution. A full competitor breakdown used to stretch across days and multiple people. Claude Code compresses that analysis down to minutes, with enough depth in the output to start acting on it the same afternoon.
The fourth pillar is where the bigger shift lives. AI citations were not a real traffic channel two years ago. Getting pulled into an AI-generated answer now produces visibility that a top-ten ranking alone cannot match, because AI Overviews and chat interfaces are taking up screen space that pushes organic results further down the page. A site sitting sixth that gets cited in the AI Overview often captures more traffic than the sites above it, a dynamic that aligns with broader findings on how AI search rewards content loops over isolated pages.
Nico kept his practical direction tight: run the four pillars in sequence. Fundamentals must come first, because any deficiencies will drag down all of the subsequent elements built upon them. The content must be created for fan-out queries as well as PAA, since that’s where the AI citations must be generated. Backlinks must be evaluated according to their quality and relevancy, and not sheer quantity. Strategy for AI citations is the final step of this process, and for a reason.
Filling up the site with articles created using AI tools in order to drive volumes, as he put it right away, leads to de-indexation rather than ranking up. It’s a well-known fact. According to a marketer Jake Ward, the content marketing strategy of his company, which was based on using an AI-powered tool that helped to generate 3.6 million visits in total from 1,800 AI-written articles aimed at Exceljet’s content, ended up with Google punishing his client, Causal, for such a strategy and causing their traffic to fall from 610,000 visits to 190,000 monthly visits.
Practical Takeaway
Pull your top competitor’s sitemap, run a fan-out query analysis on your seed keywords using a tool that separates fan-out from PAA, and check whether your competitor is cited in AI Overviews for those queries. Most are not yet. That gap is among the least contested competitive advantages sitting in organic search right now.
