Technical SEO: AI Isn't Enough Anymore—Here's What Really Matters
Technical SEO, while essential for ensuring your site is crawlable by search engine bots, is no longer enough on its own to guarantee your visibility to generative AI systems. According to Google patents such as Knowledge Vault and Knowledge-Based Trust, search engines now evaluate the accuracy of facts and the reliability of sources. I have observed, through numerous client audits, that a technically flawless website can be ignored by AI if its content is not perceived as a trustworthy source and corroborated by other entities on the web.
Case study
I recently worked with an e-commerce site specializing in eco-friendly products. Their technical SEO was exemplary: Core Web Vitals in the green, complete Schema.org markup, and optimized internal linking. Yet their products never appeared in AI-generated search results. After revising the thematic depth of their product pages—by citing scientific studies and recognized certifications, and securing mentions from authoritative blogs—their AI citation rate increased by 351% in three months. The key insight: AI doesn’t just look for information; it looks for evidence and authority.
- Flawless technical SEO has become a prerequisite, not a competitive advantage.
- Google no longer just indexes pages; it evaluates entities and the accuracy of the facts they contain.
- Google's patents and publications (Knowledge Vault, Knowledge-Based Trust) indicate a shift from exogenous signals to endogenous signals.
- Valid structured data is no guarantee of anything if your facts aren't corroborated elsewhere on the web.
- To be cited by AI, you must become a source that is selectable, consistent, and trustworthy.
You have a flawless website. Your Core Web Vitals are in the green, your Schema.org markup is complete, and your internal linking is clean. And yet, you don’t appear anywhere in the results generated by artificial intelligence. I know this frustration well; I see it often with clients in Bayonne and Bordeaux alike. We’ve done everything by the book, and the results are eluding us. So I’ll be frank with you: technical SEO alone is no longer enough. In this article, I’ll explain why—backed by Google patents—and, more importantly, what’s taking its place.
When technical SEO reaches its limits
Let’s start by setting the record straight. Technical SEO is still essential—I’m not denying that. But today, I see it more as a ticket to the game than as a trophy.
What technical SEO can do—and what it can't
Technical SEO serves one specific purpose: to make your site readable by search engine crawlers. Speed, indexability, tags, structured data, architecture—all of these facilitate crawling and indexing. It’s useful, it’s necessary, and it’s measurable. But here’s the blind spot: technical SEO optimizes the form, not the background. It tells the search engine, «Here’s a clean page.» It doesn’t say, «Here’s a reliable source on a topic it really knows well.» Yet it is precisely this second question that AI-powered search engines are now asking themselves.
The symptom, well-documented but never mentioned
You may have already noticed this. Your pages are indexed, they even rank well, and yet they don’t appear in AI Overviews or in responses from ChatGPT or Perplexity. This discrepancy isn’t a bug—it’s a sign. Generative engines no longer rank links; they compose a response based on the sources they deem most trustworthy. Even the best markup in the world won’t get you into that response if the engine doesn’t recognize you as a credible entity. And that’s where everything changes.
From page indexing to entity evaluation
To understand this shift, you need to change your perspective. Google no longer thinks in terms of «pages»; it thinks in terms of «entities.».
From keyword to entity
For twenty years, SEO was all about placing keywords in the right spots and building up backlinks. Today, Google is trying to understand what and whose You’re right. An entity is something in the real world—a brand, a place, a person, a concept—connected to others through properties. An SEO agency in Anglet, an SEO consultant in Saint-Jean-de-Luz, a freelancer in Pau—all are entities, with attributes, relationships, and a reputation. The search engine no longer asks itself just «does this page contain the keyword?»; it asks «is this entity relevant, recognized, and consistent?».
The Knowledge Graph: A Layer of Interpretation
At the heart of this system lies the Knowledge Graph, a vast network of knowledge that connects entities to one another. When you type in a query, the search engine resolves its ambiguity in real time, thanks to this graph. As an SEO consultant, my view is that this graph is the real battleground today. As long as your entity isn’t clearly and consistently represented there, you’re playing in the minor leagues. And no expert can get around this reality through simple technical optimization alone.
What Google's patents and publications reveal
I know, talking about patents might seem dry. But they tell a clear story—the story of a search engine that’s learning to judge truth, not just popularity.
The Assessment and Identification of Entities
Several Google patents describe how the search engine identifies entities, assigns them scores, and links them to other related entities. Patent US11734287B2, for example, filed in 2020 and published in 2023, details a system that identifies entities associated with a query and then ranks them by relevance. The underlying logic is clear: the search engine maps out a world of entities before even considering how to rank your pages.
Information extraction, the Knowledge Vault
Next came a major publication, «Knowledge Vault,» authored by Google researchers in 2014. The idea is mind-boggling: building a web-scale knowledge base by automatically extracting facts from text, tables, page structure, and annotations. Each extracted fact is assigned a probability of accuracy. In other words, the engine no longer simply reads your pages; it extracts facts from them and calculates whether they are likely to be true. This, in my view, is the turning point that many SEO professionals did not see coming.
Data validation: reliability backed by facts
And that’s where the centerpiece comes in: the 2015 paper «Knowledge-Based Trust.» Its proposal is radical: to evaluate the quality of a source not by exogenous signals, such as links, but by endogenous signals—namely, the factual accuracy of what it claims. A source that contains few factual errors is deemed trustworthy. The researchers applied this method to 2.8 billion facts to estimate the reliability of 119 million pages. I’ll leave it to you to consider what that implies. Your data may be perfectly tagged, but if it is false or unverifiable, it does you a disservice.
Why valid structured data is no guarantee
This brings me to the most persistent misconception in our field. «My Schema is valid, so I’m in the clear.» No, and I’ll tell you why.
Valid markup does not mean verified information
Valid markup means only one thing: your code follows the syntax. It says nothing about the accuracy of what you’re declaring. You can mark up a 4.9-star rating, a price, or a founding date—all of these can pass validation and remain unverified. However, the engine seeks to corroborate this information. Markup is a declaration, not proof.
Cross-verification from multiple sources
Verification is the key to success. The algorithm cross-checks your claims against what other online sources say. If your site reports information that no other reliable source confirms, your claim remains weak. Here’s a simple explanation of what’s at stake.
| What technical SEO optimizes | What AI-driven indexing evaluates |
|---|---|
| Page speed and indexability | Recognition of the entity in the Knowledge Graph |
| Syntax validity of the markup | Factual accuracy and data verification |
| Inbound links (exogenous signals) | Reliability of the source (endogenous signals) |
| Presence of the keyword | Consistency and authority on the subject matter |
You can see the gap. Technical SEO focuses on the left column, while AI evaluates the right column. If you want to delve deeper into the purely technical aspects of CMS, I cover that in detail elsewhere. WordPress optimizations that lay a solid foundation.
The Evolution of Algorithms: From Relevance to Trust
Let’s take a step back. This story follows a consistent trajectory, and it’s not a recent development.
E-E-A-T, authority, and entity consistency
Initially, PageRank assessed popularity based on links. Then the «Reasonable Surfer» model refined this by weighting links according to their likelihood of being clicked. Today, E-E-A-T—Experience, Expertise, Authority, and Trustworthiness—takes this logic a step further. The search engine wants to know who is speaking, with what legitimacy, and with what consistency. A consistent entity—from the website to external profiles—inspires greater trust. This applies to an SEO agency in Biarritz just as much as it does to a freelancer in Bayonne: consistency is key.
The cost of recovery
There is also a concept that is rarely discussed: the cost of retrieval. The search engine is constantly balancing the quality of a page against the computational cost of indexing and serving it. A page that is confusing, redundant, and costly to process starts at a disadvantage. In my opinion, prioritizing clarity and conciseness isn’t just a matter of convenience; it’s a factor the search engine uses to make decisions.
What's taking the place of technical SEO
Well, enough with the diagnosis—let’s talk about solutions. Because all of this leads to a concrete strategy, and I find it quite exciting.
Become a selectable source
The new goal is no longer to be ranked, but to be selected, as a source of an answer. This requires clear, structured, well-sourced content that presents a clear point of view and verifiable facts. One paragraph, one idea. One subheading, one promise. Logical transitions that guide the reader—whether human or algorithmic. This, I believe, is the best SEO investment you can make today.
Aligning the organization, from the product to CSR
There’s one tricky issue I often come across. How can a site establish itself on topics of authority, such as social responsibility, when its entire legitimacy rests on product pages? It’s a serious question, and the answer isn’t obvious. I’ve explored it in more detail here, How can you make CSR content visible when the site's topical authority is based on a product-focused approach? ?. My advice can be summed up in one sentence: don’t spread yourself too thin; tie every new topic back to your core expertise.
Frequently asked questions
Has technical SEO become obsolete?
No, absolutely not. It’s still a prerequisite—without it, nothing works. But it’s no longer enough to set you apart or get you picked up by AI. Think of it as the foundation: essential, but invisible once the house is built.
Is structured data still useful?
Yes, they help the engine understand your entities. But the fact that they are syntactically correct does not mean they are true. Make sure your facts are accurate and can be verified elsewhere on the web.
Do you need an SEO consultant or an agency to adapt?
It depends on your resources. A freelance SEO specialist or consultant is often sufficient for a small or medium-sized business, while an SEO agency is the best choice for projects involving multiple websites. The key is to choose an expert who thinks in terms of entities and reliability, not just tags.
Sources and references
- Dong, X. L. et al., «Knowledge Vault: A Web-Scale Approach to Probabilistic Knowledge Fusion», Google, 2014.
- Dong, X. L. et al., «Knowledge-Based Trust: Estimating the Trustworthiness of Web Sources», Google, 2015.
- Patent US11734287B2, identification of entities associated with a query, Google, filed in 2020, published in 2023.
- Official Google Search Central Documentation, on structured data.
Personalized, no-obligation analysis, response within 24/48 hours with 3-5 concrete quick wins.
150 entrepreneurs have already put their trust in us
🔒 Your data is never shared with third parties
