For nearly a decade, the SEO industry operated under a comfortable assumption: if you built enough content around a topic, covered every conceivable subtopic, and interlinked it all properly, Google would reward you with authority. Rankings would follow. Traffic would grow. The strategy had a name — topical authority — and it worked beautifully. Until it didn’t.
Now, as AI-generated search results from Google’s AI Overviews, ChatGPT, Perplexity, and other large language model-powered systems reshape how users find and consume information, the old playbook is proving insufficient. Not wrong, exactly. Just incomplete. And for businesses that built their entire organic strategy on topical depth alone, the reckoning is already underway.
Kevin Indig, a seasoned growth advisor and SEO strategist, laid out the case in a recent analysis for Search Engine Land. His argument is sharp and direct: topical authority remains a necessary condition for visibility, but it is no longer a sufficient one. The rules of discovery have changed because the systems doing the discovering have fundamentally changed how they evaluate, select, and surface content.
That distinction matters enormously.
Traditional search worked on a relatively straightforward retrieval model. Google’s crawlers indexed pages, evaluated them against ranking signals — backlinks, keyword relevance, site structure, domain authority — and served up a list of ten blue links. Publishers who understood this system could engineer their way to the top by producing comprehensive content clusters. Write a pillar page on “email marketing,” surround it with dozens of supporting articles on every related subtopic, and watch the domain’s authority on that subject grow in Google’s eyes. The approach was systematic, predictable, and enormously profitable for those who executed it well.
AI search doesn’t work this way.
Large language models don’t retrieve information the same way traditional search algorithms do. They synthesize. They draw from multiple sources simultaneously, weigh the credibility and specificity of claims, and generate composite answers that may or may not link back to any single source. Google’s AI Overviews, which now appear at the top of a growing percentage of search results, pull from several pages to construct a unified response. The user gets an answer. The publisher may or may not get a click.
Indig’s analysis, as reported in Search Engine Land, identifies several dimensions that AI systems evaluate beyond mere topical coverage. Among them: the specificity and originality of the information, whether the content provides genuine first-party data or unique analysis, the authority of the author (not just the domain), and how well the content answers the actual intent behind a query rather than simply matching keywords.
This is a significant shift. For years, SEO professionals could get away with what might charitably be called “content manufacturing” — producing large volumes of keyword-targeted articles that covered a topic from every angle but offered little original insight. The content was competent. It was comprehensive. And it was, in many cases, interchangeable with what a dozen other sites had published. Google’s algorithm, with its heavy reliance on backlinks and domain metrics, often rewarded these sites anyway.
AI systems are proving harder to fool.
The reason is architectural. When a large language model processes a query, it doesn’t just look at whether a page covers a topic — it evaluates the substance of what’s actually being said. Does this content contain a claim that’s well-supported and specific? Does it offer a perspective or data point that other sources don’t? Is the author someone with demonstrable expertise? These are qualitative assessments that traditional ranking algorithms struggled to make but that LLMs, trained on vast corpora of text, are increasingly capable of approximating.
Consider a practical example. A site that has published 200 articles on project management software might rank well in traditional Google search for dozens of related queries. But when a user asks ChatGPT or Google’s AI Overview for a recommendation, the system doesn’t simply count articles. It looks for the most authoritative, specific, and well-sourced answer. A single article from a respected industry analyst — someone with a verified track record — might carry more weight than an entire content cluster from a generic publisher.
This has profound implications for content strategy.
The SEO industry is already grappling with the fallout. According to data tracked by multiple analytics firms and discussed widely in industry forums, AI Overviews are appearing in roughly 30-40% of Google searches as of mid-2025, up from sporadic testing just 18 months ago. And when they appear, click-through rates to organic results drop significantly. Search Engine Land has covered this trend extensively, noting that publishers who relied heavily on informational queries — the very queries most likely to trigger an AI Overview — are seeing the steepest traffic declines.
So what’s the alternative? Indig and other strategists aren’t arguing that topical authority should be abandoned. Far from it. The argument is that it must be augmented with what some are calling “entity authority” — the idea that both the author and the publishing entity need to be recognized as genuinely authoritative by AI systems, not just by Google’s traditional link graph.
Entity authority is built differently than topical authority. It requires real-world signals: author bylines that link to verifiable professional histories, citations in other authoritative publications, appearances in structured data sources like knowledge panels, and consistent attribution across the web. When an LLM encounters a claim attributed to a named expert with a traceable body of work, it weights that claim differently than an identical claim on an anonymous blog post. The content might be the same. The trust signal isn’t.
This tracks with what Google itself has been signaling for years through its E-E-A-T framework — Experience, Expertise, Authoritativeness, and Trustworthiness. But where E-E-A-T was previously more of a quality guideline that influenced rankings indirectly (primarily through human quality rater assessments), AI systems are making these evaluations computationally. The framework has gone from aspirational to operational.
There’s another dimension that Indig highlights: the importance of what he calls “information gain.” This concept, which has roots in information theory, refers to the degree to which a piece of content adds new information beyond what’s already widely available. AI systems, by their nature, are trained on enormous quantities of existing content. They know what the consensus says. What they’re looking for — and what they tend to surface — is content that goes beyond the consensus. Original research. Proprietary data. Novel frameworks. Contrarian analysis backed by evidence.
This is terrible news for content farms. It’s excellent news for genuine subject matter experts.
The practical implications are already visible. Companies that invested in original research programs — annual surveys, proprietary benchmarking data, industry reports built on first-party information — are finding their content cited more frequently in AI-generated answers. Meanwhile, sites that built their traffic on aggregated, rewritten, or lightly differentiated content are watching their visibility erode.
And the erosion isn’t gradual. Several industry observers have noted that AI search tends to produce winner-take-most dynamics. When an AI system synthesizes an answer, it typically draws from a small number of sources — often three to five. Being one of those sources means everything. Being the sixth-best source means nothing. There is no second page of AI results.
This concentration effect is reshaping competitive dynamics across industries. In sectors like finance, healthcare, and technology — where informational queries drive enormous search volume — the stakes are particularly high. A financial services company that previously ranked on page one for hundreds of investment-related queries might find that AI Overviews now answer those queries directly, citing only the most authoritative sources. If you’re not one of those sources, your organic traffic doesn’t decline by 10%. It can decline by 80%.
Some companies are responding by doubling down on brand building. The logic is straightforward: if AI systems favor recognized authorities, then becoming a recognized authority is the most durable SEO strategy available. This means investing in thought leadership that actually leads — not the corporate blog posts that pass for thought leadership at most companies, but genuine intellectual contributions to a field. Publishing original research. Speaking at industry events. Getting cited by peers. Building the kind of reputation that an AI system can verify across multiple independent sources.
Others are focusing on structural optimization — ensuring their content is formatted in ways that AI systems can easily parse and cite. This includes clear, well-structured data markup, concise answer-format paragraphs near the top of articles, and explicit attribution of claims to named experts. It’s a technical layer that sits on top of content quality, and it matters more than many publishers realize.
But perhaps the most significant strategic shift is the move away from volume-based content strategies toward what might be called precision content. Rather than publishing 50 articles on a topic to establish topical authority, the emerging best practice is to publish fewer, better articles — each one containing original data, expert analysis, or a unique perspective that AI systems will recognize as genuinely additive.
This is a hard pill for many organizations to swallow. The content-volume approach was appealing precisely because it was scalable. You could hire freelancers, build editorial calendars around keyword research, and produce content at industrial scale. The new approach demands something different: actual expertise, original thinking, and the kind of depth that can’t be manufactured by someone who learned about a topic last Tuesday.
The irony is rich. For years, the SEO industry told businesses they needed to create “helpful, people-first content” — Google’s own language. Most businesses nodded along and then continued producing keyword-optimized content designed primarily for algorithms. Now the algorithms have gotten smart enough to tell the difference. And the businesses that actually took the “people-first” advice seriously are the ones best positioned for AI search.
There are legitimate concerns about where this trend leads. If AI systems increasingly answer queries directly — pulling from authoritative sources but not necessarily sending traffic to them — then even the best content strategy may not generate the organic traffic it once did. This is the zero-click search problem writ large, and it’s not clear that any amount of optimization can fully solve it.
Some publishers are exploring alternative models. Subscription revenue. Direct audience relationships built through email and community platforms. Partnerships with AI companies that compensate publishers for training data. These are early-stage experiments, and none has yet proven to be a reliable replacement for organic search traffic at scale.
But the strategic imperative is clear. Topical authority isn’t dead. It’s just not enough anymore. The organizations that will thrive in AI-mediated search are those that combine deep topical coverage with genuine expertise, original data, verifiable author credentials, and content that adds something new to the conversation. Everything else is noise. And AI systems are getting very good at filtering out noise.
The transition won’t be comfortable for an industry that built its practices around gaming traditional algorithms. But then, the best SEO professionals always understood that the discipline was ultimately about understanding how discovery systems work and aligning content accordingly. The discovery systems have changed. The alignment must follow.
WebProNews is an iEntry Publication