EEAT Google: Meaning, How It Works, and Practical Checklist After the October 2026 Update

EEAT stands for Experience, Expertise, Authoritativeness, and Trust: the four aspects Google uses to describe helpful and trustworthy content. EEAT is not a single score or ranking factor. It is a framework that explains what Google's ranking system tries to recognize, and what thousands of quality raters assess when examining search results.

This topic is important again because on October 1, 2026 Google revised two pages of Search Central documentation at once: the page about people-first content and the generative AI content guide. The content touches directly on common practices on blogs and SaaS sites: author bylines, profile photos, and content written with the help of AI.

This article explains what each letter means, how Google actually uses EEAT, what changed last month, and then a little checklist and script to audit your own pages.

What do each letter EEAT mean?

This concept has been around for a long time. Its initial version, EAT, appeared in quality rater guidelines since 2014. Google added the first letter E (Experience) in December 2022 to appreciate content written from direct experience, rather than simply summarizing other people's writing.

Why is Trust the most important?

Google itself states on the Creating helpful, reliable, people-first content page that of the four aspects, trust is the most important, and the other three aspects contribute to trust. Rater guidelines make a similar point: unreliable pages are rated low, no matter how experienced or expert the writer appears.

Google also explains that content is not required to show all four aspects at once. Product reviews can be strong because of experience, while technical guides can be strong because of expertise. What is non-negotiable is trust.

An easy way to imagine it: Experience, Expertise, and Authoritativeness are the pillars, Trust is the roof. Tall poles are no use if the roof leaks.

Is EEAT a ranking factor?

Not directly. Google writes that EEAT itself is not a specific ranking factor. The ranking system uses a mixture of many signals that together help identify content with good EEAT.

What about quality raters? According to Google, raters cannot determine the ranking of any page, and their rating data is not used directly in the algorithm. Google likens it to a restaurant guest feedback card: used to check whether system changes lead to better results.

So there is no "EEAT score" that you can increase with one trick. What you can do is create a page that is trustworthy, then make sure the proof of trustworthiness is clearly visible to both readers and machines.

YMYL: where EEAT weighs the most on the scales

Google gives greater weight to EEAT for topics that could have a major impact on health, financial conditions, people's safety, or the well-being of society. This category is called YMYL (Your Money or Your Life).

In the Indonesian context, for example: articles about online loans, how to report taxes, medication dosage, crypto investment, BPJS procedures, or personal data security. For topics like this, Google says content must be highly accurate and in line with expert consensus.

What changes on October 1, 2026?

On the same day, Google revised two Search Central pages. According to Search Engine Journal and Search Engine Roundtable, changes to AI content guidelines are in the official changelog, while changes to people-first pages are not recorded there and were only discovered through archive comparisons.

1. Fake author profiles are now explicitly called scams

In the "Who" section of people-first pages, Google still recommends accurate bylines. Then there's a new paragraph: creating fake creator profiles, for example using AI-generated facial photos, made-up names, or fake credentials to make the content look like it was written by a human expert, is classified as a form of deception. Google added that any fraud makes the page untrustworthy, both by users and by its automated quality systems.

This idea is actually already in the rater guidelines. What's new is the audience: it's now written directly for site owners, and mentions automated systems, not just raters. Google didn't explain how they detected it, so consider this a policy line, not evidence of a new detector.

2. Four "main content" attributes

The same page gets a new section about main content, namely the part of the page that directly helps the page achieve its goals: main text, images, videos, interactive tools such as calculators, user comments if it is the main page, tab contents, and titles and headings. Raters are trained to assess four attributes:

3. Manual fact-checking for AI content called “critical”

Generative AI content guidelines now explain that language models predict word order, rather than extracting facts, so the results can contain hallucinations. Google then wrote that it is important to fact-check and manually review all AI-generated content before publication. Search Engine Journal noted the previous version (December 10, 2025) only asked for accuracy without manual check instructions.

The coverage does not stop with the content of the article. This review also applies to metadata: title, meta description, structured data, and alt text. If a plugin or model writes these fields, they also need to be human readable.

Important to note: Google does not say AI content is penalized because it is AI-generated. Since 2023 the position has been consistent: AI content is valued like other content. The problem lies in unverified or mass-produced content with no added value.

How to use the "Who, How, Why" test?

Google offers three simple questions to evaluate its own content:

• Who: who made it? Is it clear to readers who the author is? Is there a byline in a reasonable place? Does the byline lead to a profile page explaining the author's background?
• How: how is it made? For reviews: how many products were tested, how, what were the results. For AI-assisted content: is the use of AI explained when a reasonable reader asks "how was this made?"
• Why: what is it made for? This is the most important thing. Correct answer: to help people who come directly to your site. If the goal is primarily to attract visits from search engines, that is not in line with what Google wants to reward. Using automation to manipulate rankings is a spam policy violation.

One additional test we use in the editorial: the author change test. Replace the author's name in an article with someone else's name. If there's not a single sentence that's wrong, the article hasn't shown anyone's experience. A strong article has details that only someone who has actually done it can write: numbers from self-testing, obstacles encountered, decisions made.

EEAT checklist that can be used immediately

Experience

• Include proof of real work: original screenshots, command results, photos, or numbers from your own experiments.
• Write down the obstacles encountered and how to overcome them. This is something that cannot be copied from another article.
• For code tutorials, run the code before publishing and display the original output.

Expertise and Authoritativeness

• The byline leads to the author's page containing background, topics written about, and a link to the public profile.
• Make sure each author is a real person who agreed to have their name used. Replace a stock photo or AI photo with a real photo, or delete the photo.
• Remove credentials that cannot be proven. If necessary, use a brand byline (e.g. "Editorial Team"), as a real company name is still accurate.
• Keep the site topical focused. Entering a new niche just because it looks busy with searches is a signal that Google calls a red flag.

Trust

• List sources for important claims, prioritizing primary sources.
• Show publish date and update date. Do not change the date if the contents have not changed significantly; Google calls out this practice specifically.
• Provide easy-to-find About, contact, and privacy policy pages.
• Add human fact-checking to all AI-assisted sections, including title, meta description, schema, and alt text.
• For YMYL topics, have a qualified reviewer read it before publication, and include the reviewer.

How to tag authors with structured data?

Structured data does not increase ranking by itself. Its function helps machines understand who the author is and when the page was created, provided the data is accurate. Example of JSON-LD for a blog article:

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "BlogPosting",
  "headline": "Cara Validasi NIK lewat API",
  "datePublished": "2026-11-02T09:00:00+07:00",
  "dateModified": "2026-11-02T09:00:00+07:00",
  "author": {
    "@type": "Person",
    "name": "Nama Penulis Asli",
    "url": "https://contoh.co.id/penulis/nama-penulis",
    "sameAs": ["https://www.linkedin.com/in/nama-penulis"]
  },
  "publisher": {
    "@type": "Organization",
    "name": "Contoh Indonesia",
    "url": "https://contoh.co.id"
  }
}
</script>

Remember the new rule above: schema contents must also be checked by humans. The name, profile URL, and date in JSON-LD must be the same as those seen on the page.

Small script for auditing EEAT signals

The following Node.js 18+ script takes a single URL and then checks for basic machine-readable signals: the article's JSON-LD, author name and URL, publication and update dates, and links to About and contact pages. This script does not judge the quality of the writing; it is still a human task.

// cek-eeat.mjs — cek sinyal E-E-A-T dasar di satu halaman
// Pakai: node cek-eeat.mjs https://contoh.co.id/blog/artikel
const url = process.argv[2];
const html = await (await fetch(url, { headers: { "User-Agent": "Mozilla/5.0 (cek-eeat)" } })).text();

const blocks = [...html.matchAll(/<script[^>]+type=["']application\/ld\+json["'][^>]*>([\s\S]*?)<\/script>/gi)]
  .flatMap((m) => {
    try {
      const j = JSON.parse(m[1]);
      return (Array.isArray(j) ? j : [j]).flatMap((x) => x["@graph"] ?? [x]);
    } catch { return []; }
  });

const isArticle = (b) => /Article|BlogPosting/.test([].concat(b["@type"] ?? []).join(" "));
const article = blocks.find(isArticle);
const author = article ? [].concat(article.author ?? [])[0] : undefined;

const checks = {
  "JSON-LD Article/BlogPosting ada": !!article,
  "Penulis punya nama": typeof author === "object" && !!author?.name,
  "Penulis punya url atau sameAs": !!(author?.url || author?.sameAs),
  "Ada datePublished": !!article?.datePublished,
  "Ada dateModified": !!article?.dateModified,
  "Ada link ke halaman About/Tentang": /href=["'][^"']*(about|tentang)/i.test(html),
  "Ada link kontak": /href=["'](mailto:|[^"']*(contact|kontak))/i.test(html),
};

for (const [k, v] of Object.entries(checks)) console.log(`${v ? "OK " : "CEK"}  ${k}`);

Results when we ran on a test page that intentionally didn't have dateModified and contact links:

OK   JSON-LD Article/BlogPosting ada
OK   Penulis punya nama
OK   Penulis punya url atau sameAs
OK   Ada datePublished
CEK  Ada dateModified
OK   Ada link ke halaman About/Tentang
CEK  Ada link kontak

The row labeled CEK doesn't mean your page is bad, it's just something you need to look at manually. Sites that render content with JavaScript on the client side may need to be checked via a browser because these scripts only read initial HTML.

Common mistakes about EEAT

• Thinking that bylines increase rankings. Google's Search Liaison once stated that bylines do not directly help rankings. Bylines help readers judge trustworthiness. Fake bylines, on the other hand, can now clearly be detrimental.
• Catch up on word count. Google insists it doesn't have a preferred word count. Filler paragraphs actually reduce quality.
• Updates the date without updating the content. This includes content signals created for search engines.
• Thinking listing lots of sources is enough. Attribution does not replace original efforts. An article that only summarizes ten links without any analysis of its own is still weak.
• Publish AI drafts without proofreading. After October 1, Google called manual fact-checking critical, including for metadata.

If you also want to control how your content is used to train AI models, read our guide to AI training opt-out signals in robots.txt. That setting is separate from search ranking and does not affect EEAT.

FAQs

What is the difference between EAT and EEAT?

EEAT is the latest version which adds Experience (direct experience) in December 2022. The other three aspects, namely Expertise, Authoritativeness and Trust, remain the same.

Is AI-generated content necessarily undervalued?

No. Google assesses quality, not how it is made. What is problematic is AI content that is not fact-checked, is mass-produced without added value, or is created primarily to manipulate rankings.

Am I obliged to write the label "made with AI"?

Google recommends disclosure when it is reasonable for readers to ask how the content was created, for example when automation played a major role. This is a recommendation, not a requirement for all pages.

Is it OK to use a brand byline instead of a person's name?

Yes. Google asks for accurate author information where readers expect it. Real company names remain accurate. What is prohibited is a false persona that is created to appear to be a human expert.

Which version of the quality rater guidelines applies now?

Most references mention the September 11, 2025 version (approximately 182 pages) as the applicable version. There are some blogs claiming another update in 2026, but we haven't found any official confirmation, so check the PDF directly on Google's site before citing a specific part number.

How long does it take to see the effects of EEAT improvements?

There are no exact numbers from Google. Quality changes are usually only reflected after the page is re-crawled and are often visible during core updates. Focus on improvements that are useful to readers, rather than waiting for one particular update.

Source

• Google Search Central: Creating helpful, reliable, people-first content (updated October 5, 2026)
• Google Search Central Blog: EAT gets an extra E for Experience (December 2022)
• Google Search Central Blog: Google Search's guidance about AI-generated content (February 2023)
• Search Engine Journal: Google Tells Sites To Fact-Check AI Content Before Publishing
• Search Engine Roundtable: Google: Manually Factcheck & Review AI-Generated Content For Accuracy (October 1, 2026)
• PPC Land: Google tells sites to manually factcheck all AI content before publishing
• BuildMVPFast: Google's Helpful Content Guidelines Now Call Fake Author Profiles Deception