In my latest posts, I wrote about GEO, about how to monitor whether your brand is mentioned by AI, and about Google’s soft transparency about ads made with AI. If you look closely, all three texts rest on the same foundation: a four-letter acronym that every digital marketing course has turned into a decorative checklist—E-E-A-T. Experience, Expertise, Authoritativeness, Trust.
The problem is that it turned into a checklist. People fill out an “about us” page and put a stock photo in the author’s footer, thinking that’s E-E-A-T. It isn’t. And the most recent Google guidelines update made this even clearer.
What E-E-A-T is, without the gloss of an online course
The acronym was born as E-A-T in 2014, within Google’s internal guidelines used to train human search quality evaluators. In 2022, it gained the second E, for Experience, specifically to capture something that expertise alone doesn’t cover: did you write about it because you studied it, or because you lived it? A doctor explaining a reaction to a medication has expertise. A patient describing what they felt after taking that medication has experience. Both matter, but for different reasons—and Google began to differentiate this explicitly.
Authoritativeness is about external reputation—what other independent sources say about you—not what you say about yourself. Trust is the umbrella that holds everything together: without trust, the other three pillars amount to nothing, which is why Google itself describes trust as the most important element of the acronym.
Trust is more mundane than it sounds
There’s an expectation that trust is something abstract, almost philosophical. In practice, a large part of what Google and human evaluators check is routine: is the site’s HTTPS configured correctly? Is there a real contact page with address, phone, and CNPJ—rather than just a generic form? Does the privacy policy exist and is it specific to the business, or is it copied from an automatic generator? Does the domain have history, or was it registered three months ago just to host content at scale?
None of these items, by itself, determines a ranking. Together, they form the backdrop that makes the rest of E-E-A-T be taken seriously—or not. A site with brilliant content but none of these basic legitimacy anchors is at a disadvantage against a mid-level competitor that did its institutional homework.
An update that few people in Brazil commented on
In September 2025, Google updated the Search Quality Rater Guidelines with an explicit focus on how to evaluate AI-generated content, and that version remains valid in 2026. The bar became clear: content purely generated by AI, without qualified human review and without original value, is classified as the lowest possible quality. AI-assisted content, on the other hand, can perform well—provided it’s accurate, filled with real experience, and edited by someone with genuine knowledge of the subject, according to a report by Search Engine Land.
This changes the equation for people who outsourced content production to AI without any review, assuming they’d save time. It’s not about whether you use AI or not in writing—it’s about whether the person signing the result actually has the knowledge to defend every sentence if someone asks.
What weighs more, according to whoever measures it every day
A study by NP Digital with 181 SEO specialists asked them to rate the weight of each E-E-A-T signal in rankings, on a scale from 0 to 10.
| E-E-A-T signal | Weight in rankings (0-10) |
|---|---|
| Backlinks and brand mentions | 10 |
| In-depth knowledge of the topic | 9 |
| Positive reviews and social signals | 9 |
| Author credentials | 8 |
| Avoid misleading content | 8 |
| Content transparency | 6 |
Notice that the top of the list isn’t something that can be fixed in a single text: it’s backlinks, brand mentions, in-depth knowledge. In other words, the signals that carry the most weight are precisely the ones that require time and consistent work to build—not a plugin or a formatting technique.
Weber had already explained why this was going to happen
Max Weber described three forms of legitimate authority: traditional (you lead because it has always been like that), charismatic (you lead because people believe in you personally), and rational-legal (you lead because you follow a formal, verifiable set of rules that anyone can audit). Digital marketing was born almost entirely as charismatic: the brand that shouted the loudest, with the most memorable slogan, earned trust through sheer personality performance.
E-E-A-T is Google trying to convert charismatic authority into rational-legal authority—by force. Instead of trusting how trustworthy a brand looks, the system tries to formalize auditable criteria: who wrote this, what that person published previously, who else cites that domain, and how long it’s been around. It’s bureaucracy in the purest Weberian sense—and like all bureaucracy, it favors whoever can document the process, not whoever only knows how to perform trust.
What the numbers say about AI citations
According to a study by RankAISearch, websites with complete trust signals have a 67% higher chance of appearing in responses generated by AI, and content with combined E-E-A-T signals appears 3.2 times more in AI search tool responses than content without those markers. Content with detailed author credentials is recommended by AI platforms 2.4 times more than anonymous content. There’s also a warning to keep in mind: these figures are the company’s own research, without independent external verification to the extent I found, so I treat them as directional, not as law.
A complementary data point from GoodFirms reinforces something I already touched on in the monitoring post: only 14% of companies today track their own visibility in AI searches. The same study shows that original content delivers 66% of ROI, compared to 42.6% for updating existing content and 42.3% for technical corrections. Producing something genuinely new is, in practice, nearly double the value of revisiting what already exists.
Where most people get it wrong
The most common mistake I see in a client site audit isn’t a lack of content—it’s content signed by no one, defended by no one. A blog page without an author name, or with an author name that’s just a generic WordPress user. A bio that says “digital marketing specialist” with no link to proof that this person exists outside that site. A customer testimonial without full name, without a company, without a way to verify it. Each of these details, by itself, seems small. Together, they build exactly the opposite of trust.
How to build real E-E-A-T, step by step
A practical process to turn E-E-A-T from an empty checklist into a real signal of trust, focusing on what truly moves the weights identified by NP Digital’s research.
- Sign every piece of content with a real, verifiable person
Full name, real photo, and a link to LinkedIn or a public professional profile. If the person signing the content doesn’t exist outside that site, the trust signal doesn’t exist.
- Write from real experience, not just research
Whenever possible, include what you yourself have lived, tested, or implemented—not only what you read somewhere else. That’s the difference between expertise and experience that Google began to distinguish since 2022.
- Invest in original content before updating what already exists
Your own data, real case studies, and research conducted internally. According to GoodFirms, original content delivers almost double the ROI of simply updating existing content.
- Pursue mentions and backlinks from sources you don’t control
It’s the highest-weight signal according to NP Digital’s research. Press, partners, class associations, and clients who are willing to publicly talk about the work.
- Review all AI-assisted content with someone who masters the subject
The September 2025 Google guidelines update is clear: AI content without qualified human review falls to the lowest possible quality classification.
What I think about it
Of the five steps above, the third is what bothers me the most—because we still haven’t solved it internally. Superplural has gathered real insight into the Brazilian market over years of campaign work, and much of that has never turned into original published content—it’s just become an internal client report. That’s exactly the kind of asset that builds real E-E-A-T, and that we’re sitting on without using. I’ll come back to this point soon, because there’s a specific study we already should have published and we still haven’t.
Besides that, what remains from this entire cluster of posts—GEO, monitoring, ad transparency, E-E-A-T—is a single message, repeated from different angles: there is no technical shortcut that replaces truly earning real trust. Technology changed the verification mechanism, but it didn’t change what needs to be true for someone to trust you.
