- Google has grown sharply less tolerant of scaled and outdated content since December 2025, with even small and low-volume sites now caught in spam updates.
- Ahrefs’ 2026 AI visibility correlation study puts branded web mentions at a 0.664 correlation with AI Overview appearances, while page count shows almost no meaningful relationship at roughly 0.194.
The Build in Public podcast episode dated July 31, 2026 opened with a sharp backstory. The guest, known in SEO circles as the Godfather of Topical Authority – Koray Tugberk Gubur, described starting in black-hat iGaming, managing 3,000 sites inside a private blog network before losing every single one of them on August 1, 2018, the day the industry later named the Medic Update. “Every August something bad happens,” he said. “I’m scared again.” Joining him was Charles, a recurring guest known for dissecting Google’s system-level behavior.
Google’s Tolerance for Scaled and Stale Content Is Shrinking Fast
The central claim: since December 2025, Google has become noticeably less forgiving toward two specific site types. Those carrying outdated content and those inflating page counts through high-volume AI publishing.
The case example used was BrowserStack. This particular website was said to have about 23,000 referring domains, which were mainly clean and authoritative sites. They created over 2,000 AI-generated articles in one day, but the site had reportedly lost nearly 70 percent of its traffic. Not portrayed as a regular penalty, but rather, as an indication that authority will no longer protect a website against scaled content penalties.
“They are trying to clean these two by being more aggressive.” According to Google’s December 2024 core update which ended on December 18, 2024, the new standards for E-E-A-T and freshness signals have been set.
Brand Mentions and Backlinks Are Doing Different Jobs Now
The sharpest exchange concerned what brand mentions actually do inside Google’s systems. Both guests agreed: unlinked brand mentions are not a substitute for a backlink. “The mention is not going to be passing the PageRank,” Gubur said. “It won’t be providing a higher rate of crawl. It won’t be impacting the freshness signals.”
The distinction drawn was between page ranking and entity ranking. Links dominate page-level signals; mentions shape how Google and AI systems recognize and position a brand in the knowledge graph. “When you try to rank your site, links are more important,” he said. “But if you are trying to be mentioned by Google as the top five consultants or something like that, mentions might be impacting the consensus faster.”
The numbers behind that distinction are hard to dismiss. Ahrefs’ 2026 AI visibility correlation study found that branded web mentions carry a 0.664 correlation with AI Overview appearances, YouTube mentions push that to 0.737, and branded anchors cross into strong correlation territory at 0.628 for AI Mode. Page count, the metric that drove enormous content budgets for years, registers almost nothing at around 0.194.
A case study made the entity logic concrete. A woman being defamed online had her knowledge panel changed entirely through semantically structured statements published across local sites, no backlinks involved. “Links were not even in the subject,” he said.
Google’s Ranking Algorithm Is Becoming a Retrieval Layer for AI
Charles framed this inside a larger structural shift. The ranking algorithm, he said, is transitioning into a retrieval and sourcing mechanism for AI Mode and AI Overviews rather than a destination in itself. “The blue links algorithm is becoming less based on the traditional need of the algorithm and more of a retrieval and sourcing algorithm for the end AI overview and AI mode.”
Research published by position.digital in July 2026 confirmed that AI Mode and AI Overviews cite the same URLs only 13.7 percent of the time, despite sharing Gemini infrastructure. Only 2 percent of cited URLs appear across AI Overviews, ChatGPT, and Perplexity simultaneously, a number that exposes how badly fractured the “AI search” surface actually is.
Exposure Ninja’s research into AI Mode illustrates the same point. Their client The Ordinary surfaces consistently across AI tools for value-oriented skincare queries, not because of any single optimized page, but because the brand concepts co-occur so frequently in indexed content that the association is embedded in the model itself.
Content Freshness Is a Scoring System, Not Just a Signal
The episode addressed freshness in terms most practitioners skip. A Google patent linked to engineers including Amit Singhal and Jeffrey Dean covers information retrieval based on historical data; Google reportedly stores up to eight prior versions of a document to score a page’s quality trajectory, not just its current state. Charles connected this directly to AI infrastructure: the algorithm needs fresh pages to generate quality answers. Freshness has shifted from a relevance signal to a sourcing requirement.
Three Workstreams, Running at the Same Time
Rankings remain the upstream filter: OptimizeGEO’s 2026 guide documents that Google AI Overviews cite top-10 organic results 76.1 percent of the time. Brand signals need deliberate engineering; Airops 2025 research cited by Goodfirms found brands are 6.5 times more likely to be cited in AI answers through third-party sources than through their own domains. And freshness needs a measurement system: pages older than six months with data no longer matching current consensus are liabilities, not neutral assets.
Semrush’s AI Visibility Index, tracking 126 million prompts across 22 industries, puts 89.3 percent of AI-search demand in categories with no dominant brand owner today. Search Engine Land has noted that topical coverage earns eligibility for citation; brand authority determines whether a brand gets chosen. Most teams are building the first and skipping the second.
What Industry Experts Are Saying?
This podcast operates in a much larger conversation among experts, one which has become increasingly more defined by 2026. Lily Ray, who serves as VP of SEO and AI Search for Amsive, has claimed that AI search visibility is essentially just E-E-A-T done well, not something different from search visibility that requires its own playbook. At Affiliate Summit West 2026, Ray was clear: “AI Google Overviews mean we have the largest AI answer surface online, clicks will keep dropping and brand visibility will be our main KPI.” The larger context Ray provides, of how AI is augmenting but not replacing search, carries an important implication: good organic rankings are still the filter.
Aleyda Solis, founder of Orainti, has pressed on a different and underappreciated gap. Her AI search optimization roadmap starts not with content structure but with infrastructure. AI crawlers behave differently from Googlebot. They hit aggressive rate limits, fail on JavaScript-heavy rendering, and are blocked outright by Cloudflare rules that many sites have never reviewed. Content optimized for citation cannot be cited if the bot cannot reach the page. That gap between content quality and technical accessibility is where many optimization efforts quietly fail.
The tension between these two positions, Ray’s emphasis on authority signals and Solis’s emphasis on technical access, is not a contradiction. They are sequential problems. A site that ranks well and builds genuine brand signals across third-party sources still loses citation potential if its infrastructure blocks AI crawlers at the door. The debate itself reflects how early the field still is.
This is part of a larger question the industry has not fully settled: whether SEO and GEO are genuinely separate disciplines or the same work measured differently. That tension has been examined closely, with evidence suggesting the framing of SEO versus GEO may itself be built on a flawed assumption. The Build in Public episode, with its emphasis on brand signals, freshness, and entity architecture, lands closer to the view that the underlying mechanics are continuous, not separate.
What the episode adds to that conversation is enforcement history. The guest has watched Google’s pattern across more than a decade, from the Medic Update erasing 3,000 sites in a single day to the current pressure on AI-scaled content at any authority tier. His read: this is not a temporary correction. It is the same enforcement logic applied to a web generating content at a pace Google has never had to manage before. The brands that make themselves easy to find, verify, and cite across surfaces are the ones built for what the algorithm needs now. Everything else is friction; it will keep filtering out.
