E-E-A-T stands for experience, expertise, authoritativeness and trustworthiness. It is the vocabulary Google gives to the contractors who score its search results, and it describes the qualities a page is expected to display before it deserves to rank. The acronym names no setting, exposes no number in any Google tool, and cannot be switched on. It describes a target that ranking systems are built to approximate.

The framework exists because credibility is not directly measurable by machine. Algorithms count links, clicks, text patterns and site history; none of those is trust. Google's answer has been to hire people to read pages against a written manual, then use their aggregated judgements to check whether changes to the ranking systems moved results in the right direction. The documentation is blunt about the limits of that arrangement: search raters have no control over how pages rank, and rater data is not used directly in the ranking algorithms. Google compares the feedback to comment cards collected by a restaurant.

How the four components fit together

The components are not equal. Google's guidance states that trust is the most important of them, with the other three contributing to it, and that content does not have to demonstrate all of them to be useful. A forum post can be helpful through experience alone. A regulatory explainer can be helpful through expertise alone.

Experience covers first-hand involvement: having used the product, visited the place, or lived through the situation. Expertise covers demonstrable knowledge, whether formal credentials or evident command of a subject. Authoritativeness covers external recognition, meaning whether others in the field treat the creator or site as a reference point. Trustworthiness covers accuracy, honesty, safety and reliability, assessed at page, site and entity level.

Weighting shifts with subject matter. Google applies stricter expectations to what it calls Your Money or Your Life topics, defined as content that could significantly affect health, financial stability or safety, or the welfare of society. Medical, financial, legal and civic content sits inside that category, and pages there are held to standards that a hobby blog is not.

The rating work itself runs through the Search Quality Rater Guidelines, a public PDF hosted by Google. Raters apply two separate scales. Page Quality runs from Lowest to Highest and assesses whether a page achieves its purpose well. Needs Met assesses how far a result satisfies the query behind it. E-E-A-T feeds the first scale rather than the second, which is why a highly relevant page can still be rated poorly.

Who, how and why

Google translated the rater vocabulary into three creator-facing questions, published on its helpful content documentation and last revised on December 10, 2025. Who created the content is answered through bylines and author pages that carry background on the people involved. How the content was created covers methodology, including the number of products tested, the evidence of that testing, and any disclosure of automation or generative systems used in production. Why the content was created is, in Google's framing, the most important of the three: content made primarily to help people aligns with what the ranking systems reward, and content made primarily to attract search visits does not.

That third question carries a policy consequence. Using automation to generate content for the primary purpose of manipulating rankings is classified as scaled content abuse under Google's spam policies, which makes it an enforcement matter rather than a quality preference.

Origin and evolution

Versions of the rater manual circulated unofficially for years before Google acknowledged them. The E-A-T formulation, without the experience component, entered the guidelines in 2014, though some accounts place its first appearance in the 2013 revision; Google has never published a definitive first date. The first officially released public version arrived on November 19, 2015.

The July 2018 revision pushed the concept further into the document, replacing generic quality language with E-A-T framing. Weeks later, the August 1, 2018 core update produced heavy movement among health and medical sites, and the search industry began treating the acronym as an operational concern rather than rater trivia.

Experience was added on December 15, 2022. Google's announcement described situations where first-hand life experience is what a searcher values most, offering the example of someone choosing between tax software and wanting a forum discussion rather than an accountant's explainer.

Revisions since then have tracked the arrival of generative systems. A November 2023 update simplified the Needs Met definitions and added guidance for forums and short-form video. In January 2025 the document grew from 170 to 181 pages, with new material on spam identification and an explicit list of the criteria raters weigh, trust among them. That April, raters were directed to treat pages whose main content is automated and adds no value as lowest quality, and a definition of filler content was introduced. According to Search Engine Land, the September 11, 2025 revision added evaluation examples for AI Overviews and renamed the Society category to Government, Civics and Society; Google described the changes as minor.

Why it matters to marketing teams

The clearest recent demonstration came from telehealth. In March 2026, Hims lifted keyword rankings 34% in three weeks with no new content and no link building, and the single largest driver, according to the practitioner who ran the work, was replacing collective editorial attributions with named physicians and their credentials. Visible editorial standards and structured medical review accounted for the rest.

That result landed against a bleak backdrop for health publishers. The December 2025 core update cut deeply into medical information sites that had held dominant positions for years, and analysis of the same update found nearly 15% of top-ten pages displaced, with regulated verticals anchored to long-established domains.

The framework has also migrated into AI surfaces. In May 2026, Google added a dedicated first-hand perspectives section to AI Mode and AI Overviews, surfacing forum and creator content alongside generated answers. One practitioner commenting on the change argued that experience now decides whether a page is cited at all, not merely where it ranks. An earlier optimisation checklist for AI search placed expert bylines and third-party mentions among the strongest inclusion signals.

Practitioner opinion tracks the same direction. A survey of 131 specialists scored authority and trust at 36.5% and brand signals at 27.0% among the factors judged to matter most, on 13,665 data points collected across more than 100 candidate signals.

Byline integrity is the weak point. Press Gazette identified crypto writers whose existence could not be verified, publishing under bylines that pass every visible authorship test. Brand safety taxonomies do not check whether an author is real.

Limitations and disputes

The central dispute concerns what the framework actually is. Google states plainly that E-E-A-T is not a ranking factor, while adding that the systems use a mix of factors capable of identifying content that displays it. Internal documentation leaked in May 2024 complicated that position. The Content Warehouse material, covering roughly 2,500 modules and 14,000 attributes, included fields named siteAuthority, isAuthor and isPublisher, alongside an author reputation score. Google confirmed the documents were authentic while cautioning against conclusions drawn from incomplete context, and the presence of a field in an API specification does not prove it is used in production ranking.

A second problem is diagnostic. No tool reports an E-E-A-T value, which means any explanation of a traffic decline in those terms is inference. That gap has consequences: after the September 2023 helpful content update, independent publishers reported catastrophic losses, with one site falling from roughly 4,000 daily visitors to about 200. At a Mountain View summit on October 29, 2024, Google acknowledged that its systems could not explain why some quality sites had been suppressed. Participants later described the exercise as a pacification tactic that produced no remedy.

A third criticism concerns incumbency. If authoritativeness is measured through external recognition, established institutions accumulate it faster than new entrants, and evaluation at entity level compounds the advantage.

Not the same as

The Search Quality Rater Guidelines are the document; E-E-A-T is one concept inside it, sitting alongside the Page Quality and Needs Met scales.

The helpful content system was an algorithmic ranking component, folded into the core systems during 2024. E-E-A-T describes qualities; the system acted on them.

YMYL classifies subject matter by potential for harm. It sets how strictly the framework is applied, not what the framework measures.

Domain authority is a vendor metric produced by third-party SEO tools. Google has consistently denied publishing an equivalent, which is why the leaked siteAuthority field attracted attention.

Author Central is Amazon's console for claiming book listings, unrelated to search authorship despite the shared vocabulary.

Recent developments

Google published a consolidated guide to optimising for generative features in May 2026, which restated existing practice rather than introducing new requirements, a position the company's VP of Search repeated to marketing leaders by framing good SEO as sufficient preparation for AI search. The third spam update of 2026 rolled out globally on August 18, continuing enforcement against scaled and low-effort content. Measurement of the AI surfaces where these signals now operate remains partial: Search Console's generative AI performance report reached every site worldwide on August 31, 2026 without click or query data.

Timeline

  • 2013 to 2014: E-A-T enters the Search Quality Rating Guidelines; sources differ on which revision carried it first
  • November 19, 2015: Google publishes the guidelines openly for the first time
  • July 2018: A revision replaces generic quality language with E-A-T framing throughout the document
  • August 1, 2018: A core update produces heavy movement among health and medical sites, pushing E-A-T into mainstream practice
  • August 25, 2022: The helpful content update targets content written primarily for search engines
  • December 15, 2022: Experience is added, creating E-E-A-T
  • February 8, 2023: Google's guidance on AI-generated content directs creators to E-E-A-T and the Who, How, Why questions
  • November 16, 2023: Needs Met definitions are simplified and guidance for forums and short-form video is added
  • March 5, 2024: A revision accompanies the March 2024 spam update
  • May 2024: Content Warehouse documentation leaks, exposing siteAuthority, isAuthor and author reputation fields
  • May 29, 2024: Google confirms the leaked documents are authentic
  • October 29, 2024: Google hosts a summit with 20 affected publishers in Mountain View
  • January 23, 2025: The guidelines expand from 170 to 181 pages with added spam material
  • April 9, 2025: Raters are instructed to rate unedited automated main content as lowest quality
  • September 11, 2025: AI Overview evaluation examples are added and the Society category is renamed Government, Civics and Society
  • December 10, 2025: The helpful content documentation receives its most recent revision
  • March 2026: A named-byline programme is credited with a 34% keyword ranking lift in telehealth
  • May 2026: First-hand perspectives becomes a dedicated section inside AI Mode and AI Overviews
  • August 18, 2026: The third spam update of 2026 rolls out globally

Summary

Who. Google defines the framework and applies it through contracted search quality raters, who score sample results rather than individual sites. Publishers, content teams, health and finance marketers, and anyone competing for citation inside AI answers are the parties affected by how it is applied.

What. A four-part description of content credibility covering experience, expertise, authoritativeness and trustworthiness, with trust treated as the element the other three support. It is not a ranking factor, not a score, and not visible in any Google reporting surface.

When. E-A-T entered the rater guidelines around 2014, went public on November 19, 2015, gained its second E on December 15, 2022, and has been revised repeatedly since, most recently on September 11, 2025 for the guidelines and December 10, 2025 for the creator-facing documentation.

Where. In the Search Quality Rater Guidelines PDF, in Google's helpful content documentation, and in the ranking and citation outcomes those documents describe, including AI Overviews and AI Mode.

Why. Machines cannot measure credibility directly, so Google built a human vocabulary for it and trains automated systems to approximate the judgements it produces. The framework matters commercially because the weighting rises sharply for topics touching health, money and safety, and because the same qualities now govern which sources generative answers cite.