E-E-A-T signals: what counts and what is myth

Sooner or later the acronym turns up in your inbox. An agency offers "E-E-A-T optimization", a plugin promises an author box "for better rankings", and a blog post claims Google has just introduced new rules against AI content. You run a physio practice, a small shop or an accounting firm, and you wonder whether you are missing something. The honest answer: part of it matters, part of it is a business built on uncertainty, and part of it is simply made up.

This piece sorts it out: where E-E-A-T comes from, how Google itself talks about it, which signals are actually meant and in what order to tackle them. For the details it points to three more articles in this series.

E-E-A-T comes from a handbook for human reviewers

Google employs people who rate search results, known as quality raters. They work from a thick handbook, the Search Quality Rater Guidelines. The current version is dated 11 September 2025 and runs to 182 pages, and the original PDF is public. It describes how raters should recognize whether a page is good.

For a long time the acronym in there was E-A-T: expertise, authoritativeness, trust. The full handbook was first published in November 2015. In December 2022 a second E was added, for experience, meaning first-hand experience. Since then it has been E-E-A-T. The latest revision in September 2025 mainly sharpened the definitions for particularly sensitive topics and added examples involving AI Overviews.

One point matters for everything that follows: the raters do not decide where your page ranks. According to Google they have no control over rankings, and their ratings are not used directly in the ranking algorithms. Google compares them to feedback cards in a restaurant: the ratings show whether search is working well overall and help evaluate changes to the systems.

The handbook is still worth reading, because it describes which pages Google wants to see at the top. According to industry reports, Google's John Mueller called it insightful in May 2026, but explicitly not a guide to ranking. Read it as a description of the goal, not a blueprint.

Four letters, and one of them carries the rest

The four terms sound abstract. In the handbook they come with simple questions and examples.

The authoritativeness example deserves a second look. Authority here does not mean "lots of backlinks" or "a famous brand". A bakery's Instagram account is the definitive source for whether there is plum cake today. A small site can perfectly well be the go-to source for its own field.

Trust sits at the center. The handbook puts it bluntly: if a page is untrustworthy for any reason, it has low E-E-A-T, however experienced or expert it may seem. Google's own example is a scammer who knows a great deal about scams: that expertise does not make him trustworthy. The other three letters feed into trust, and a page does not need to show all three equally. How much trust is needed depends on the purpose: a product review has to be honest, a page about medication or taxes has to be accurate, a funny post about Mondays needs almost none of it.

No score, no single ranking factor, but not irrelevant either

This is where most of the confusion lives, because two true statements seem to contradict each other.

Statement one: E-E-A-T is not a ranking factor. Google's documentation on helpful content says E-E-A-T itself isn't a specific ranking factor. There is no value Google calculates for your site and stores somewhere. According to industry reports, Google's Gary Illyes said back in 2019 that there is no internal E-A-T score, and Danny Sullivan repeated in 2024: not a ranking factor, not a score, not an algorithm.

Statement two: Google's systems still look for signals connected to it. The same documentation says the systems use a mix of factors to identify content with good E-E-A-T. For topics that could significantly affect people's health, financial stability or safety, or the well-being of society, such content gets even more weight. Sullivan wrote in 2019, in essence: if by ranking factor you mean something directly measurable like page speed, E-A-T is not one. If you mean that many signals act as proxies for it, then in that sense it is.

The two statements fit together. E-E-A-T describes what Google wants to find. The algorithms do not measure E-E-A-T; they measure many individual things that tend to go along with trustworthy pages. Google names one of them publicly: whether other well-known sites link to or mention the content. Most of the others it does not name.

Sullivan summed up what this means for you in 2024: instead of trying to prove E-E-A-T, ask whether visitors arriving from search are satisfied with what they find. That is less comfortable than a checklist, but it is the question all the signals lead back to.

The May 2024 leak of internal Google documentation does not change much here. It shows that Google stores authors and a kind of site authority as data. Whether and how strongly that feeds into ranking it does not say, and the term E-E-A-T does not appear in it.

And because health, money and legal matters keep coming up: the handbook calls these topics YMYL, "Your Money or Your Life". Whether your site is affected is decided not by your industry but by the individual page. That is a topic of its own, covered in detail in YMYL pages.

What "signals" actually means: three layers

The handbook names three sources from which E-E-A-T can be read: what the site or the person says about themselves, what others say about them, and what is visible on the page itself. That gives three layers, which differ in how much of it you control.

On the page: the content itself

Here you control everything. The most important signal is the content: does it answer the question, is it correct, did real work go into it? The handbook names effort, originality, talent or skill, and accuracy as yardsticks, and for sensitive topics accuracy is measured against expert consensus.

Then experience. Google asks whether content shows first-hand knowledge, for instance because someone actually used a product or actually visited a place. On a small site that means your own photos instead of stock images, before and after measurements, a concrete case with a timeframe and an outcome, a sentence on how you tested and where the limits are. The fictional example-physio.com does not write "back pain can have many causes"; it shows which three exercises it guides patients through in the first weeks after a slipped disc, with its own photos from the treatment room.

Add the visible evidence: a byline that leads to more information about the author, sources for figures and claims (primary sources where possible), and a visible date with a label, "published" or "updated". The date has to be honest. Google's own self-assessment questions explicitly ask whether you change dates to make pages seem fresh when the content has not substantially changed. Why that backfires is explained in updating existing content.

On the site: who is behind it

The second layer surrounds every page. The handbook expects it to be clear who is responsible for the site and who created the content, and that visitors can get in touch. How much of that is needed again depends on the purpose: for sites that handle money the handbook calls contact information extremely important, and for shops raters specifically look for payment, exchange and return policies.

In practice that means an about page with real names and real photos instead of "we are a young, dynamic team", a contact page with an address and phone number, author pages for everyone who writes advice content, and a recognizable purpose for the site. If you operate in Germany, Austria or Switzerland, a legal notice (Impressum) is mandatory anyway, and it is exactly what a rater looks for under "who is responsible". That is a pointer, not legal advice; check the details with your chamber of commerce, professional body or a lawyer.

How strict the handbook is here shows in one line from the criteria for the lowest rating: a complete lack of information about who is responsible for the site and its content, on pages that require trust. How to set up bylines, author pages and an about page is covered in author pages.

Beyond the site: what others say about you

The third layer is the one you control least, and the handbook still weighs it heavily. Raters are required to research reputation for every rating, with searches like example-accounting -site:example-accounting.com. What counts are independent reviews, press articles, forum mentions, recommendations from experts. When self-description and independent sources disagree, the independent sources win.

Three details from the handbook help here. For sensitive topics, what experts in the field say is what counts, and a recommendation from a professional association is strong evidence. Customer reviews should be read, not just counted: an unfriendly employee mentioned in a review is not a problem, credible reports of fraud are. And traffic statistics from third-party tools are explicitly not evidence of reputation.

What you can do is manageable: a well-kept business profile, asking satisfied customers for a review without incentives, making memberships in professional bodies or associations visible, contributing articles or talks where your professional audience reads. Bought reviews are off limits, both legally in many countries and under Google's policies. The full list of these signals, including what matters for shops, is in website trust signals.

Who, how, why: Google's three questions for any piece of content

Written for site owners is Google's guide to helpful content. It breaks E-E-A-T down into three questions.

Who created it? Can readers see without hunting who wrote the content, and does the byline lead to more information about the person? Google explicitly encourages accurate authorship information where readers would expect it. Nobody asks that on a "boiler servicing" service page. Everyone asks it on a guide to property tax.

How was it created? For a review: how many products, with what results, with what evidence? For automated or AI-assisted content: is that apparent, and is it explained how and why it was used? An explanation is worth adding where readers would wonder about that.

Why does it exist? Google calls this perhaps the most important question. The content should primarily help people, not primarily attract visitors from search. An article that exists only because a keyword tool reported search volume fails here, however tidy its byline.

AI content: origin does not decide, added value does

Google's position has been the same since February 2023: quality is rewarded however it is produced, and using AI gives content no special advantage. Using AI to manipulate rankings violates the spam policies.

The handbook is clear in both directions. Since January 2025 it defines generative AI as a helpful tool that, like any tool, can be misused. Many low-effort pages with no value for visitors get the lowest rating, explicitly regardless of how they were created. At the same time it states that the use of AI alone says nothing about a page's effort or quality.

One example from the handbook shows how not to do it: a parenting guide whose terms of service vaguely mention that "some articles" are AI-generated, without saying which, and which describe the site as an experiment for AI enthusiasts rather than a resource for parents. It gets the lowest rating.

For you this means: AI as a tool for outlining, drafting or phrasing is fine, as long as a person with expertise checks it, adds to it and takes responsibility for it. First-hand experience, your own photo, the concrete case from your practice are things a language model cannot supply, and they are exactly what sets one page apart from a hundred similar ones.

Small site, no press coverage: less of a problem than it sounds

No Wikipedia entry, no press articles, twelve Google reviews: how is that supposed to compete with big portals?

The handbook has a section on exactly this, and it is reassuring. Small sites often have little or no reputation information online. That, it says, is not indicative of high or low quality. Raters should then rely more on the other criteria, especially for sensitive topics. Elsewhere it states that "High" and even "Highest" ratings are possible without any reputation information.

So a lack of fame is not a flaw, a bad reputation is. A handful of honest reviews and a site that shows who does the work and what they know is a solid position for a local practice. Do not inflate anything artificially. The handbook explicitly warns against exaggerated self-promotion: E-E-A-T should rest on the content, reputation research and verifiable qualifications, not on the claim "I'm an expert". Overdoing it lands a page at "Low".

Five myths that cost time and money

"There is an E-E-A-T score." There is not, neither public nor internal, and several Google employees have repeated this over the years. Anyone selling you an E-E-A-T value has invented their own number. It may work as a checklist, but it does not measure what Google sees.

"An author box brings rankings." Google recommends bylines for readers. On author pages, Mueller said in 2020, in essence, that they are not a must but make sense if you want people to trust the information. A box with an invented title actually hurts: the handbook treats fabricated profiles and false qualifications as deception.

"Schema markup creates authority." Structured data describes what is on the page and helps Google identify authors unambiguously. Mueller said in 2025, in essence, that structured data won't make a site rank better. Marking up things readers cannot see violates Google's guidelines. What markup does and does not do in the age of AI answers is covered in structured data for AI.

"Backlinks equal authority equal E-E-A-T." Half true. Google names links and mentions from well-known sites as one factor. But authoritativeness in the handbook means whether someone is the go-to source for a topic, and trust weighs more. Metrics like Domain Authority or Domain Rating come from tool vendors, not from Google.

"In June 2026 Google tightened the guidelines." Several blogs report an update on 12 June 2026 introducing a new flag called "Synthetic Authority" against AI content. None of them links to a Google source, the section numbers they cite do not match the real handbook, and as of 1 October 2026 the official PDF still carries the date of 11 September 2025. This is how myths spread: one post makes a claim, others copy it. When someone argues with a new rule, ask for the link to the PDF or the Google blog.

The order: trust first, then evidence, then markup

Many people start with a schema plugin, which is the wrong end. A better order is the one in which a skeptical visitor experiences your site.

First, the foundations of trust. Legal notice where required, contact details, an address, an about page with people on it, HTTPS, and for shops, findable information on shipping, returns and payment methods. Without this layer everything else is a façade.

Then the content with the highest risk. The pages where a mistake could harm someone: are the facts right, are sources given, does a named person stand behind it? A wrong tax allowance on example-accounting.com weighs more than ten missing author pages.

Then make experience visible. Your own photos, figures and cases on the pages that currently sound interchangeable. This is where small businesses have an edge over big portals.

Then bylines, author pages and honest dates, linked cleanly from the byline to the author page and on to the about page.

Markup comes last, as a mirror of what is visible, not a substitute for it. Reputation is something you tend on the side, continuously.

Where to start

1. Search for your brand with yourname -site:your-domain.com and read the first two results pages the way a stranger would.
2. Check that the legal notice and contact page are directly reachable from every page, and that your about page shows real names and photos.
3. List the five pages where a mistake could harm a visitor, and check facts, sources and dates there.
4. Add a visible byline to advice content that leads to an author page with a short, truthful bio.
5. On three pages, replace stock photos and generic statements with your own images, measurements or a concrete case.
6. Check that the displayed modified date only changes when the content actually changes.
7. Only then: add Article, Organization and ProfilePage markup and check it with the Rich Results Test.

How Traffalyzer helps

First things first: Traffalyzer cannot measure E-E-A-T, because no such value exists. The scan checks the technical groundwork the visible signals build on.

It counts how many of your pages carry author information in their metadata (meta name="author" or article:author) and how many have a published or modified date (article:published_time, article:modified_time). According to its own documentation, Google does not use either author tag; what counts there is the visible byline and the author in Article markup. The number is therefore a pointer for other systems such as AI assistants, feed readers and social networks, and it shows whether authors are being maintained at all. The scan also measures coverage with Schema.org structured data and checks HTTPS.

It flags modified dates that are not credible: a page reports a fresh date, but Traffalyzer has seen no change to the content in weeks. The typical cause is a theme that sets the date anew on every page view. The "Content to refresh" list suggests pages that are starting to go stale, and with Search Console connected you can see whether rankings rise after a revision. Whether your bio is accurate and your photos are real is still for a person to judge.

Frequently asked questions

What does E-E-A-T stand for?

E-E-A-T stands for experience, expertise, authoritativeness and trust. The term comes from Google's guidelines for human quality raters. It describes how to recognize a reliable page, with trust at the center and the other three contributing to it.

Is E-E-A-T a ranking factor?

Not in the sense of a single, measurable factor. Google says E-E-A-T itself isn't a specific ranking factor, and Google employees have repeatedly said there is no score for it. Its systems do use a mix of signals connected to E-E-A-T, though, and weigh them more heavily for topics involving health, money and safety.

Do quality raters rate my website?

They rate samples of search results, but their judgments do not change where any individual page ranks. Google uses the ratings to check whether changes to its algorithms improve search overall. No penalty follows from it.

Do I need author pages for E-E-A-T?

Not everywhere. Bylines and author pages are worth it where readers would ask who wrote this, especially for advice on health, money or legal matters. A service page does not need one. Google has never confirmed author pages as a direct ranking lever; their value lies in readers' trust.

Can a small site without press coverage or a Wikipedia entry have good E-E-A-T?

Yes. The guidelines explicitly say that little reputation information online is, for small sites, neither a sign of high nor of low quality. Content, experience and transparency then count for more. What hurts is not an unknown name but a bad reputation.

Is AI content downgraded because of E-E-A-T?

Not because of its origin. Google assesses quality regardless of how content was produced. What gets downgraded is mass-produced pages without added value, whether written by people or machines. AI used as a tool, with expert review and your own experience added, is not a problem.