E-E-A-T: how to show it on your site and what's known about its effect

E-E-A-T stands for experience, expertise, authoritativeness and trust, the criteria Google uses to describe reliable content. The full definition is in what is E-E-A-T. This guide covers something else: how to show it on your site, using what Google publishes, and what's really known about its effect. The awkward answer comes first: Google says it isn't a specific ranking factor, and we found no public measurement that isolates its effect.

What Google says, and what it doesn't

The reference is Google's guide to helpful content, updated on 5 October 2026. It says four things worth reading together:

• E-E-A-T itself isn't a specific ranking factor, but Google's systems use a mix of factors that identify content with good E-E-A-T.
• Of the four aspects, trust is the most important. The others contribute to it, and content doesn't have to demonstrate all of them.
• Those systems give even more weight to E-E-A-T on topics that could affect people's health, financial stability or safety (YMYL).
• It asks you to consider who made the content, how and why, and calls "why" perhaps the most important question.

Google doesn't publish which factors these are, how much each one weighs, or any score you can look up. So be wary of any figure like "good E-E-A-T lifts you X positions": we know of no primary source behind one.

The closest thing to a signal: Q*

One document says a little more. The US case against Google produced, with redactions, notes of a call on 18 February 2025 with Google engineer HJ Kim (exhibit PXR0356). According to those notes, "Q* (page quality (i.e., the notion of trustworthiness)) is incredibly important". They add that it's largely static, related mainly to the site rather than the query, and that PageRank is one of its inputs.

It's easy to read too much into this. These are notes of a call, not sworn testimony or technical documentation, and much of what makes up Q* is redacted. The document doesn't mention E-E-A-T, so it doesn't show that Q* is E-E-A-T, or that an author byline moves it. It does support one useful idea: reliability is also assessed at site level, and one page with good signals won't fix an unreliable site. That reading is ours.

What quality raters do, and what they don't

Nobody at Google is scoring your site to rank it. According to Google, quality raters have no control over how pages rank, and their data isn't used directly in its ranking algorithms. Google compares them to feedback cards in a restaurant, and says they're used to benchmark results and test changes before launch.

Their handbook is useful to you, though. The rater guidelines (11 September 2025 version, the one Google was serving on 6 November 2026) ask raters to judge E-E-A-T using one or more of these three things:

• What the site or creator says about themselves: the About page or profile.
• What others say about them: independent reviews, references, news articles.
• What's visible on the page: the main content, reviews and comments.

The second is the one you control least and the one most often forgotten. There's also a warning that rarely gets quoted: a clear conflict of interest reduces trust. The handbook's examples are the manufacturer's own "review" and the paid influencer's. If your site compares products you sell, say so.

Start with trust

The guidelines define trust as the extent to which a page is accurate, honest, safe and reliable, and add that an untrustworthy page has low E-E-A-T however experienced it seems. What's required depends on the type of page:

Page type	What the guidelines ask for	Where to look
Online store	Secure payment and reliable customer service	Checkout, contact details, returns terms
Product review	Honest and written to help people decide, not solely to sell	Whether you say what you tested and what interest you have
Information on a clear YMYL topic	Accurate, to prevent harm	Who writes it and which sources it cites
Social post on a non-YMYL topic	May not need a high level of trust	Nothing special

The third column is our reading, as the guidelines give no such list. The practical order follows from this: before adding bios, fix what's false or out of date, and make it visible who's responsible for the site.

How to show you've done it

Saying "we have experience" proves nothing. Google gives more concrete pointers: with reviews, it helps to explain how many products were tested, what the results were and how the tests were run, with evidence of the work such as photographs. Its generative AI guide contrasts a first-hand review with a summary of what already exists.

Two examples of ours, public and checkable, in our comparison of Google Analytics alternatives:

• Our own measurement, with its limits. We downloaded each tool's script, noted the bytes, explained how, and said what we didn't measure.
• An admitted error. The article says the previous version promised 30 tools, listed 32 and described fewer.

The rule we take from this is simple: a claim of experience should be checkable. A bare client count tells readers nothing, while a case with a period, method and source does. If you can't publish client data, show the method with an example and say it's an example. What you should never do is make it up. Google calls fabricated profiles and false credentials deception, and the guidelines require the lowest rating for invented author profiles or ones that claim a qualification the person doesn't hold.

Our example: what we do on zds.es

Here's what we apply as this article goes out, and you can check each point, for instance in the glossary entry on E-E-A-T:

• A visible byline leading to an author page, also in the article's structured data.
• A visible review date in the glossary.
• Sources linked at the end of each glossary entry, and in newer entries the date they were consulted.
• In this article and in the analytics comparison, a closing paragraph on how they were made, including the use of AI.

What we don't know: we haven't measured whether any of this moves rankings or citations, so we don't claim it does. We do it because Google asks for it and because it's useful to a reader.

Authors and structured data

Structured data helps Google work out who's who. It doesn't make anyone an expert. What the documentation says:

• Article author: include in the markup every author shown on the page, each in their own field. Link to a page that identifies them, using url or sameAs.
• Author profile: if it's a page on your site, use ProfilePage.
• Organization: Organization has no required properties, so add those that apply. There's more in our Schema.org guide.
• Reviews of your own business: marking them up won't get you stars. Since 2019 Google hasn't shown those results for LocalBusiness and Organization when the entity being reviewed controls the reviews, including through a third-party widget. Ask for them on outside platforms, which also count as "what others say".

Is your topic YMYL?

The guidelines decide this by possible harm, treat it as a spectrum, and note that many topics aren't YMYL. Three of their examples: buying prescription drugs is clearly YMYL, a car review may be, and buying pencils isn't. They suggest two questions: would a careful person seek out experts to prevent harm? Would most people be content asking their friends?

The financial category refers to what could damage a person's ability to support themselves and their family. The guidelines don't treat B2B as a category, and we didn't find the word in them. An article about business software isn't YMYL just because the software is expensive. On a YMYL topic, experience and expertise aren't interchangeable either: personal experience is accepted in a forum about coping with treatment, while treatment options and filling out tax forms are left to experts.

If you lost traffic in a core update

A drop after a core update doesn't mean your E-E-A-T fell, and Google gives no such diagnosis. Its guide (10 December 2025) asks for this:

1. Wait until the update has finished and at least a full week more. In Search Console, compare that week with one before the rollout began.
2. Look at your top pages and queries. Google's example of a small drop is position 2 to 4, where nothing drastic is needed. Its example of a large one is 4 to 29.
3. If the drop is large and sustained across the site, go through its self-assessment questions. You can ask someone unconnected with the site to do it.
4. Avoid quick fixes. Deleting content is a last resort.

What Google calls a small drop and a large one

Small: position 2 to 4

Nothing drastic needed

Large: position 4 to 29

If sustained across the site: self-assessment questions

Position 1

30

These are examples from Google’s guide, not limits.

Google warns that some changes show in days and others take several months, and that there's no guarantee of recovering positions. There's more context in our article on core updates.

Does it help in AI answers?

We don't know, and we found no primary source that shows it. Google's generative AI guide (10 July 2026) doesn't mention E-E-A-T once: we counted in the HTML on 6 November 2026. That's a fact about that page and doesn't show it has no influence. What it does say about features such as AI Overviews:

• What will matter most in the long run is unique, useful content with its own point of view and first-hand experience.
• There's no special Schema.org markup to add for generative AI search.
• Seeking inauthentic "mentions" doesn't help as much as it may seem.
• Search Console's generative AI report lets you measure performance in these features.

We don't cover ChatGPT, Perplexity and other engines here, as we haven't read any documentation of theirs on E-E-A-T.

What to do tomorrow

1. Fix what's false or out of date on your most visited pages.
2. Check that each article has a real byline that leads to a page about that person.
3. Look for any credential or figure you can't back up. If you can't, remove it.
4. In one important article, add how it was made: what you tested, which sources you read and what role AI played.
5. Declare conflicts of interest in comparisons.
6. Mark up author and organization, and stop marking up reviews of your own business.
7. Decide whether your topic is YMYL using the two questions above. If it is, have an expert write or review it.
8. Search for your brand away from your site and read what people say about you.

FAQ

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

Google says E-E-A-T itself isn't a specific ranking factor, though it uses a mix of factors that identify content with good E-E-A-T. It doesn't publish which ones or how much they weigh.

Is there an E-E-A-T score I can look up?

Not in the Google documentation we read. If a tool gives you a figure, that's the tool's own calculation.

How long does it take to show?

Google gives no timeframe for E-E-A-T. For improvements after a core update, it says some changes show in days and others can take several months, with no guarantee.

Can I use AI to write?

The guidelines say using generative AI doesn't by itself determine the level of effort or quality: it can produce good and bad content. What they mark down is producing pages in bulk with little effort and faking human authors with invented profiles. Google also suggests explaining how AI was used if it generated much of the content.

Does every page need a named author?

Google recommends bylines where readers would expect them. A product page or contact page usually won't have one, and what counts there is who's responsible for the site.

Sources

Retrieved on 6 November 2026.

• Google: creating helpful, reliable, people-first content, 5 October 2026
• Google: search quality rater guidelines, 11 September 2025
• Google: how Search is tested
• Google: guide to generative AI features on Search, 10 July 2026
• Google: core updates, 10 December 2025
• Google: Article, ProfilePage, Organization, September 2026
• Google: review snippets, 8 September 2026, and announcement on self-serving reviews, 16 September 2019
• Google: the extra E for experience, 15 December 2022
• Exhibit PXR0356: notes of the call with HJ Kim, 18 February 2025, redacted public version, in the Department of Justice list

How this article was made

The previous version claimed that E-E-A-T matters more than ever, that high E-E-A-T makes an AI more likely to cite you and that all B2B content is YMYL, without sources. Those statements are gone. This text was rewritten in November 2026 with AI assistance, from the pages linked above, and reviewed before publication. We haven't run our own experiment on the effect of E-E-A-T.