Google can't call your university to verify your degree. It can't check whether you've treated patients, built bridges, or managed ad campaigns. It doesn't know if you're a real expert or someone who's very good at looking like one.
What it can do is check the web's receipts. And it's getting much better at it.
The March 2024 core update introduced three new spam categories that directly target manufactured authority: scaled content abuse, site reputation abuse, and expired domain abuse. That wasn't a minor policy tweak. It was Google telling the market: we know what you're doing, and we're building systems to catch it faster.
How Google verifies authority without verifying anything
Google's E-E-A-T framework describes what good content looks like. But E-E-A-T isn't a score. There's no API endpoint that returns your authority number. Google evaluates authority through proxy signals, indirect evidence that a person or site has genuine expertise.
The proxy signals fall into a few categories:
Who links to you. Not how many links you have, but who's doing the linking. A citation from a university research paper carries different weight than a link from a site you own. Google's link graph analysis has been running for over two decades, and it's the foundation of how they distinguish earned authority from constructed authority.
Who mentions you. Entity recognition lets Google track mentions of people, brands, and organisations across the web, even without links. If journalists, industry publications, and conference organisers mention you in contexts you don't control, that's a signal. If you only appear on properties you own or pay for, that's a different signal.
Where you've been published. A byline on an established publication with editorial standards tells Google something different than a self-published post on a site registered last month. The publication's own authority factors in.
Whether your claims are consistent. If you claim to be a cardiologist on your website but no medical institution, medical publication, or professional directory mentions you as one, that inconsistency is detectable. Google cross-references entity data across the Knowledge Graph, Wikipedia, professional databases, and structured data across the web.
"Top 15 Financial Advisors in the Southwest, 2024." The publication looks like a regional business magazine. It isn't. The domain was bought six months earlier for $200, loaded with filler articles, and dressed up with a $50 theme. The financial advisor at #3
E-E-A-T is one of the most misunderstood concepts in SEO. People treat it like a checklist they can tick off or a score they can optimize for. It is neither. Understanding what E-E-A-T actually is, and what it is not, makes the difference between chasing phant
ChatGPT for SEO strategy means using the model to accelerate the research, analysis, and planning stages of SEO, not to replace the strategic thinking that makes a plan worth executing. You can build a complete quarterly SEO plan in a few hours instead of a fe
None of these signals is definitive on its own. Google uses them in combination. The question has always been: how good is Google at distinguishing the real thing from a convincing imitation?
Earned authority vs manufactured authority
The distinction is straightforward in theory. Earned authority comes from doing work that other people independently recognise. Manufactured authority comes from creating the appearance of recognition yourself.
The practical challenge is that manufactured authority, done well, looks exactly like earned authority from the outside. Someone who publishes a list naming themselves as a top expert, then cites that list on other properties they control, then gets that claim scraped by AI systems, has created a chain of apparent validation. We've covered that specific playbook in detail.
Google's job is to untangle those chains. And the gap between earned and manufactured authority is exactly where Google has been focusing its engineering resources. Every major algorithm update of the past few years has tightened the proxy signals, made cross-referencing harder to fake, or added new categories of manipulation to the detection pipeline.
The March 2024 core update was the biggest step yet.
The March 2024 core update: three new weapons
Before March 2024, Google's spam policies covered the obvious stuff: keyword stuffing, cloaking, doorway pages, link schemes. These are first-generation manipulation tactics. The March 2024 update added three new categories that target more sophisticated approaches.
Scaled content abuse
This replaces Google's earlier, narrower "auto-generated content" policy. The old policy focused on whether content was machine-generated. The new policy focuses on whether content, however it's produced, exists primarily to manipulate rankings.
The shift matters. It means Google isn't asking "did AI write this?" anymore. They're asking "was this created to help users or to game search results?" A human can write 500 thin articles targeting long-tail keywords, and it's just as much scaled content abuse as having a bot do it.
In practice, this targets content farms that pump out hundreds of pages with minimal editorial oversight, whether they use AI, freelance writers, or a combination. The signal Google looks for is pattern: similar structure, similar depth (or lack of it), similar keyword targeting, all pointing to a revenue-capture strategy rather than genuine information sharing.
Site reputation abuse
This is the one that sent shockwaves through affiliate marketing. Google now specifically targets sites that host third-party content to exploit the host site's ranking signals.
The classic example: a news site with strong domain authority hosts a "Best Credit Cards of 2026" page written by an affiliate network. The page has nothing to do with the news site's editorial mission. It exists because the affiliate team knows the news domain will rank it higher than their own site would.
Google calls this "parasitic SEO" internally, and the March 2024 update gave them formal policy ground to act on it. Sites received manual actions for hosting coupon pages, product review sections, and sponsored content that was clearly designed to piggyback on the host's authority rather than serve the host's audience.
Expired domain abuse
Buying an expired domain with existing authority and repurposing it to rank new content isn't new. What's new is Google formally classifying it as spam. The update targets cases where someone acquires a domain primarily for its backlink profile, then fills it with unrelated content designed to rank on the strength of those old links.
This is a direct attack on a cornerstone of manufactured authority. Expired domains were the raw material for private blog networks for years. Google's detection of PBNs has improved steadily, but the expired domain policy gives them a broader enforcement tool that covers even standalone repurposing.
SpamBrain: how Google's detection learns
SpamBrain is Google's AI-based spam detection system. Google first mentioned it publicly around 2018, and it's been a central part of their anti-spam infrastructure since.
What makes SpamBrain different from rule-based spam detection: it learns patterns. Rather than relying on engineers writing rules like "flag sites with more than X% exact-match anchor text," SpamBrain identifies spam by recognising patterns across Google's entire index.
This feedback loop is why manufactured authority has a shrinking shelf life. Every spam technique that works today becomes training data for tomorrow's detection. The link networks that slipped through in 2023 were used to train the models that caught them in 2024.
Google's Webspam Report for 2024 claimed SpamBrain detected over 5 billion spam pages per day. Even discounting for marketing inflation, the scale of automated detection means that any technique used widely enough to be profitable becomes visible enough to be detected.
SpamBrain is the engine. But knowing how the engine works is less useful than knowing what it catches reliably and where it still falls short.
What Google detects well right now
Some manufactured authority tactics are now reliably caught.
Private blog networks. PBNs were the workhorse of manufactured authority for a decade. Google's detection is now strong enough that PBNs have a short effective lifespan. SpamBrain identifies shared hosting patterns, similar site templates, unnatural link timing, and the characteristic footprint of sites that exist only to pass link equity. Some PBNs still work briefly, but the average time-to-detection has shrunk from years to months. We've written a full breakdown of how link rings work and how to detect them.
Link spam networks. Paid link schemes, link exchanges, and coordinated linking patterns are caught with increasing reliability. The December 2022 Link Spam Update and subsequent updates have progressively narrowed the window for link-based manipulation. Google's stated approach is to neutralise spam links (ignore them) rather than penalise the target site, but persistent link spam can still trigger manual actions.
Scaled AI content. Mass-produced AI content with minimal editing follows detectable patterns. Not because Google can identify AI writing per se, but because scaled content tends to be structurally uniform, topically shallow, and optimised for keywords rather than user intent. The content itself might be grammatically flawless, but the publishing pattern gives it away.
Parasite SEO. Since the March 2024 update, Google has issued manual actions against prominent sites hosting third-party content for ranking manipulation. Several major publications had their coupon and affiliate sections deindexed. This was the highest-profile enforcement action in years, and it signalled that even high-authority domains aren't immune when they're hosting content that doesn't serve their audience.
Expired domain repurposing. Google's ability to detect when a domain's content has been completely replaced while its backlink profile remains from the previous owner has improved substantially. The mismatch between historical link context and current content is a strong signal.
What Google still struggles with
This is where it gets more interesting for anyone thinking about the long-term trajectory of search quality.
Sophisticated multi-brand networks. When one person or entity controls a personal brand site, an agency site, a niche blog, a review site, and a podcast, and all of them cross-reference each other, the network can look completely natural. Each property has real content. Each property has its own design and voice. The linking between them looks like what you'd expect from a professional with multiple legitimate ventures.
Google's challenge: how do you distinguish this from an actual professional who legitimately runs multiple web properties? The signals overlap almost completely. Unless the cross-linking is blatant or the content is thin, these networks can persist for years.
Self-published authority claims. Creating a "Top 10 Experts" list, ranking yourself #1, and citing it across your other properties works because Google treats mentions and citations as authority signals without reliably tracing them back to their origin. The system sees "this person is mentioned as an expert on five different sites" without always recognising that all five sites are controlled by the same entity.
This is the tactic that AI systems amplify most. When an AI model scrapes a self-published list and presents it as fact in a search response, the manufactured authority gets laundered into a new medium with no editorial filter.
Conflict of interest masking. Someone who runs an SEO agency, moderates a major industry community, and publishes "independent" tool reviews has a structural conflict of interest. When that community control extends to coordinated subreddit takeovers, the scope of influence becomes even harder for algorithms to map. But conflict of interest isn't a spam signal. Google's systems don't evaluate whether a content creator has financial incentives that might bias their content. They evaluate whether the content itself appears authoritative.
This gap is conceptual, not technical. Google has stated repeatedly that they want to reward helpful content regardless of who creates it. But when the creator's helpfulness is shaped by undisclosed commercial interests, the content can look helpful while serving the creator more than the audience. The problem compounds when Reddit content surfaces directly in search results, because Google treats community-moderated discussions as authentic signals without evaluating who controls that moderation.
Gradual authority building. The hardest manufactured authority to detect is the kind that's built slowly, with genuine effort, but with manipulative intent. Someone who spends two years building real content, earning some legitimate links, and gradually introducing self-serving citations is nearly invisible to algorithmic detection. The signals look organic because, at the surface level, they are.
The verification gap and why it's closing
Google's fundamental problem is that expertise is a human judgment. You can verify credentials against databases, but you can't algorithmically determine whether someone understands their subject. The proxy signals are the best available approximation, and they're getting better for three reasons.
The data set is growing. Every manual action, every confirmed spam report, every algorithmically detected manipulation becomes training data. SpamBrain's model improves with each iteration. The techniques that work today become the patterns detected tomorrow.
Cross-property analysis is improving. Google has been investing in entity-level evaluation, understanding not just individual pages but the relationships between sites, brands, and people across the web. WHOIS data, hosting patterns, content fingerprints, link graphs, and entity co-occurrence data all contribute to mapping who controls what.
AI Overviews raise the stakes. When Google surfaces information in AI Overviews, they're not just ranking a page. They're making a factual claim with their own brand behind it. That increases their incentive to verify the authority of their sources. A manipulated ranking result is embarrassing. A manipulated AI Overview citation is a liability. The same dynamic plays out with AI encyclopedias being gamed by people who understand how training data flows from web content to model output.
The practical result: the window between discovering a new authority manipulation tactic and Google detecting it at scale is shrinking. PBNs worked for five-plus years before reliable detection. Parasite SEO worked for maybe two years at scale before the March 2024 crackdown. The next generation of tactics will likely have an even shorter effective lifespan.
What this means if you're building real authority
The sections above lay out what Google can detect, what still slips through, and how fast the gap is closing. Given those capabilities and limitations, the practical question is: how do you position yourself on the right side of that line?
If you're doing it right, Google's improvements are unambiguously good news.
Every algorithm update that catches manufactured authority makes genuine authority more valuable. When a competitor who's been gaming expert lists or running a link network gets caught, the sites with real signals benefit. The playing field tilts toward people who've invested in actual expertise and earned independent recognition.
Practically, here's what matters:
Diversify your citation sources. Don't rely on a small number of properties you control. Genuine authority shows up as mentions and links from sources you had no hand in creating. Pitch to publications. Speak at events. Contribute to research. The harder a citation is to manufacture, the more Google values it.
Build topical depth over time. A site with 200 thin posts targeting keywords looks different, at the data level, from a site with 50 deep articles demonstrating progressive expertise on a subject. Google's content quality models have gotten better at recognising depth versus breadth-without-depth. Tools like Ooty's Content Analyzer can help you identify where your content has genuine depth versus where it's surface-level coverage.
Make your credentials verifiable. If you claim expertise, make sure it's confirmable through sources you don't control. A professional association listing, a conference speaker bio, a university affiliation page, these corroborate your claims in ways Google can cross-reference.
Watch your link profile. Earned links from diverse, independent sources are the strongest authority signal. If your link profile is dominated by self-controlled properties, guest posts on low-editorial-standard sites, or reciprocal arrangements, your authority signals will look manufactured even if they aren't. Run your domain through a schema validator to make sure your structured data accurately represents your authority claims.
Be patient. Genuine authority takes years to build. That's a feature, not a bug. The time investment is what makes it hard to fake. If someone offers you a shortcut to authority, they're offering you something Google is actively building systems to detect and devalue.
The direction of travel
Google's spam detection isn't just getting better at catching today's tactics. It's getting better at catching the pattern of manipulation, regardless of the specific technique.
That's the real takeaway from the March 2024 update. The three new spam categories aren't just three new rules. They represent a shift toward evaluating intent behind content and authority signals, not just their surface appearance.
The gap between manufactured and genuine authority is narrowing from Google's side. It won't close completely. There'll always be sophisticated operators who stay ahead of detection for a while. But the cost of maintaining manufactured authority is rising, the lifespan of new tactics is shrinking, and the reward for genuine expertise is increasing.
For anyone building a long-term presence in search, that's the trend that matters. Build something real. The algorithms are getting better at telling the difference.
