TL;DR

The purpose of EEAT is to attempt to maintain the reliability and usefulness of search. Experience is based on first-hand, personal interaction with a topic (e.g. a patient’s review), Expertise is based on formal knowledge and skill (e.g. a dentist’s medical advice), Authority is an assessment of the usefulness of a person’s observations based on their expertise, experience and importance to domain knowledge.

You can demonstrate these signals through experiential and (appropriate) technical language, formal accreditation, external mentions from trusted domains, and establishing a long-term record of authority in a niche – and prove your EEAT through citations, referencing, comprehensive schema and through accurate platform profiles.

Introduction

In the last couple of months, I’ve covered how search is developing, I’ve talked about why the ‘great EEATification’ of search is important and why it has to happen, and now I’m going to give you the information you need to lead the way for your industry vertical. Some of them are pretty simple, some of them take time (some of them we’re still working on ourselves, because the cobbler’s children are always barefoot), but with EEAT and its component parts set to continue growing in importance, starting now can give you an important head start over your competitors – however search develops.

What is EEAT (last time, we promise)?

This isn’t for SEO, we promise – you can tell because we’ve consciously begun the first sentence without the keyword. Feel reassured? I hope so. In the TL;DR section of our last blog on the subject, I explain the purpose of EEAT like this:

The easy mass production of content by AI has caused a severe crisis of trust in search results, leading users to seek “real information” elsewhere (e.g. appending ‘reddit’ to searches). EEAT (Experience, Expertise, Authority, and Trustworthiness) is the essential, proven solution. The techniques that successfully established trust for high-stakes YMYL topics must be strengthened and extended across all search areas to allow Google to regain user trust and act as a reliable source of information.

While this is the purpose, what it doesn’t explain is exactly what the acronym means, so that’s what we’re going to tackle here:

Experience

Consider the extent to which the content creator has the necessary first-hand or life experience for the topic. Many types of pages are trustworthy and achieve their purpose well when created by people with a wealth of personal experience. For example, which would you trust: a product review from someone who has personally used the product or a “review” by someone who has not?

• Google’s Quality Evaluator Guidelines

Added to Google’s Search Quality Rater Guidelines in December of 2022, experience is described in the linked Search Central article as follows:

Does content also demonstrate that it was produced with some degree of experience, such as with actual use of a product, having actually visited a place or communicating what a person experienced? There are some situations where really what you value most is content produced by someone who has first-hand, life experience on the topic at hand.

For example, if you’re looking for information on how to correctly fill out your tax returns, that’s probably a situation where you want to see content produced by an expert in the field of accounting. But if you’re looking for reviews of a tax preparation software, you might be looking for a different kind of information—maybe it’s a forum discussion from people who have experience with different services.

This is a common practice when describing both of the Es of EEAT – they tend to be described relationally to one another: expertise is this, because experience is this, and vice versa. There’s a good table a little further below, stolen reproduced from the Search Quality Evaluator Guidelines which makes the relational nature of the two Es nice and clear, but in the simplest possible terms I can think of at the time of writing – if you want to know how good a dentist is, you go to his patients and query their experience, if you want to know what an appropriate dental procedure would be for a given issue, you would rely on that dentist’s expertise.

What we’re never likely to get from Google, however, is how their various algorithms determine either experience or expertise before the results are ready to be evaluated (partly because they don’t know themselves thanks to the black-box nature of the multi-layered neural networks that make up many of their most recent algorithms), but we can draw inferences and make rational assessments of the data most likely and most widely available to the algorithms. 

We can then use those assessments to make recommendations, then the impact of those recommendations to correlate them with positive and negative outcomes. We’re sure we’re not the only one who’s brain will automatically say ‘correlation is not causation’ whenever they see that word – but it’s impossible in many cases to give direct ‘causative’ advice (beyond links, really), so we have to rely on our experience here (that’s a point, right, QRG reviewer?).

Language and co-ocurrence

The most obvious source of data for algorithms to use to judge experience is the language on the page. That means that you will need to use language which can, without any question, be linked to a genuine experience of or with a subject or object. That includes common linguistic links to experience such as the following:

Verbs (actions / processes of experience)

• Used
• Tried
• Tested
• Built
• Experiencing
• Feeling
• Sensing
• Perceiving
• Noticing
• Observing
• Encountering
• Undergoing

EXAMPLE: After encountering this product for the first time, and feeling the smooth finish of the casing, it was not long before I started noticing…

EXPLANATION: Obviously, using all of these words in a row makes for some pretty poor reading, but when dealing with experiential verbs, you’re looking at words like these and related to these – words which clearly express interaction with something.

Nouns (states or instances of experience)

• Experience
• Sensation
• Perception
• Emotion
• Awareness
• Encounter

EXPLANATION: As with experiential verbs, you’re looking at nouns (naming) words and related words which clearly express the actual use or application of something.

Adjectives (describing the nature of experience)

• Experiential
• Sensory
• Emotional
• Cognitive
• Perceptual
• Immersive

EXPLANATION: Adjectives, as noun modifiers, are great for communicating experience – what colour is a thing, how big is the topic, how broad is the selection. As with the other words listed, you’re looking at words and related words clearly expressing use.

Friction and edge-case language (issue related language)

• Struggled
• Ran into
• Had issues with
• Figured out
• Got around
• Learned the hard way

EXPLANATION: Language which describes friction is as clear an indication of use or experience of a product or thing – this is language which doesn’t need to be negative, but does need to show a barrier climbed, so to speak – a difficult concept to grasp, but which unlocked a new understanding, perhaps, or a button that was a little stiff on first use, but which proved to be a temporary issue.

Comparative language (opinion language)

• Better than
• Worse than
• Compared to
• Prefer
• Outperforms

EXPLANATION: Commonly used in reviews, this kind of language implies a reference class – which usually, in turn, implies experience.

Co-occurence

Obviously, use of these words alone is insufficient – algorithms won’t find one of these words and instantly understand that the writer has experience, but what it can is make mathematical inferences from things like co-occurrence – how often such language appears together and how often using those as experience signals results in positive feedback during reinforcement learning. 

Again, because modern machine learning has moved to more opaque learning systems (RankBrain, apparently, originally primarily used supervised learning), we can’t say anything definitively, but we can make educated human judgements of the available materials, and co-occurrence of experiential language elements would be an intuitive starting point for any assessment of experience.

External indicators

These are slightly more complex – but won’t be news to anyone with a little experience of SEO. External indicators of experience boil down to graph based or reputational signals which directly or indirectly support experience.

Links and mentions

Linked and unlinked mentions have been part of SEO best practice for some time (though we’re increasingly seeing newly minted experts state unlinked mentions are somehow unique to AIO/GEO), and, in short, are reputational indicators from external publications. This doesn’t mean that any link or mention will do, however. Though links have always been strongly correlated with search performance, what you’re looking for are links from highly relevant, trusted domains. 

That doesn’t mean looking at DA or equivalent metrics (though they can be useful), it means looking for sites that are active, trusted in their own right (active forums, good estimated traffic numbers, good visibility)  and relevant to the topic at hand. Links and mentions – especially descriptive (with descriptive text around the link or mention) – are excellent external indicators of your experience.

Community signals

Whether forums, social media or other sites in a similar niche, mentions and (again) links from a community that deals with similar subjects/objects are also excellent indications of your experience.

Entity creation

This has, historically, been boiled down to simply creating an ‘author’ page, or adding some author schema to a blog post, but entity creation is about not only the entity itself, but also how it interacts with a broader network of entities over time. A single profile is a good start, but how well reproduced is that entity-level information across platforms, how long has this entity been connected to this particular topic or domain, how well is it referenced by relevant third parties, what new knowledge does the entity provide the domain?

Relational explanation of experience vs. expertise

                                                                                                                                                                                                     

YMYL Topic	Valuable sharing of life experience	Information or advice best left to experts
Sleep challenges when pregnant (YMYL Health or Safety)	Safe and non-medical tips and tricks for sleeping in the last trimester of pregnancy, provided by people who have personally struggled with this challenge, for example: how to use pillows to sleep comfortably in a position that is safe for babies	Sleep medications that are safe during pregnancy
Liver cancer treatment (YMYL Health or Safety)	A sincere and respectful forum discussion where people are describing how they’re coping with liver cancer treatment	Different treatment options for liver cancer and the associated life expectancies under each treatment
Filling out tax forms (YMYL Financial Security)	A humorous video from a non-expert content creator about the frustration of doing taxes	Instructions on how to fill out tax forms
Saving for retirement (YMYL Financial Security)	Reviews of retirement saving services by people with first-hand experience using these services	Advice on how to invest for retirement: how much to save, what kinds of assets to invest in, how much money you will need to retire by a specific age, etc.
How to vote (YMYL Society)	A social media post by an ordinary citizen describing why they personally believe it’s important to vote in local elections	Information about who is eligible to vote or how to register to vote

Expertise

Consider the extent to which the content creator has the necessary knowledge or skill for the topic. Different topics require different levels and types of expertise to be trustworthy. For example, which would you trust: home electrical rewiring advice from a skilled electrician or from an antique homes enthusiast who has no knowledge of electrical wiring?

• Google’s Quality Evaluator Guidelines

No doubt, as you were reading the experience section, you’ll have been saying ‘yeah, but this is applicable to all aspects of EEAT’ – and you’d be right. The general methods of proving EEAT tend to be similar across each of the individual letters, so we’re not going to repeat ourselves for the sake of it. Instead, we’ll focus on where proving expertise differs from the proof of experience. External links, community mentions, profile consistency etc. all matter to demonstrating expertise, but where expertise differs are the following:

Formal accreditation

As with the dentist example given earlier, the difference between the experiencer and the expert is often formal – you can experience something without the need of any domain level knowledge. 

You can experience drunkenness, for example, without a formal qualification from the Institute of Brewing and Distilling (IBD), but we know we’d much prefer our drink was brewed by someone who did. For that reason, formal accreditation is an excellent way to demonstrate expertise – and many fields provide an online directory of accredited professionals which can be linked to in a [sameAs] field in an individual’s schema. 

Semantic and linguistic analysis

Similar to the ‘language and co-occurrence’ section, demonstration of expertise can be achieved linguistically (on the page) – but with slight differences. Domains become breeding grounds for unique jargon – we’re not talking useless business jargon, here, which is often used in an exclusionary and/or obfuscatory manner, but jargon in its traditional sense:

“specialized terminology associated with a particular field or area of activity.^[1] Jargon is normally employed in a particular communicative context and may not be well understood outside that context. The context is usually a particular occupation (that is, a certain trade, profession, vernacular or academic field), but any ingroup can have jargon. The key characteristic that distinguishes jargon from the rest of a language is its specialized vocabulary, which includes terms and definitions of words that are unique to the context, and terms used in a narrower and more exact sense than when used in colloquial language.”

While good communicators will try to reduce the impact of jargon on the reader – explaining unique or rare examples, industry specific communication – even when trying to be as clear as possible – will often feature such linguistic shorthand. A short trip through this blog will turn up examples of SERPs, ranking factor, EEAT, SEO/GEO/AIO and many more which we use (often without thinking about it) to reference broader concepts, or simply for ease (not having to write search engine results pages every time has probably saved me literal hours over the years).

These semantic and linguistic patterns are easy to recognise – and appropriate usage (co-occurrence with other known terminology) is also an observable pattern. So while I wouldn’t recommend using jargon for the sake of it, using it in service to clear, industry specific communication, is a clear indication of expertise.

Domain interaction and referencing

This is also a really good trustworthiness signal, so we’ll go a little deeper there, but domain experts will interact with their knowledge domain in subtly different ways than non-experts. For example, were I to attempt to explain some subtle aspect of physics, we’d probably have to refer to a science communicator – Brian Cox (not that one), or Carl Sagan, while a genuine domain expert would likely refer to recent academic publications.

Authority

Consider the extent to which the content creator or the website is known as a go-to source

for the topic. While most topics do not have one official, Authoritative website or content creator, when they do, that website or content creator is often among the most reliable and trustworthy sources. For example, a local business profile page on social media may be the authoritative and trusted source for what is on sale now. The official government page for getting a passport is the unique, official, and authoritative source for passport renewal.

• Google’s Quality Evaluator Guidelines

As stated in the expertise section, I’m only going to add the additional stuff here – so you can safely assume that proving authority is an ‘all of the above and’ situation rather than just needing the following.

Longitudinal data

One unique method of algorithmically assessing the authority of an entity to discuss a subject/object is simply the length of time they’ve been doing so. While you may not be able to assess whether they’re a good or bad authority on the topic, a 20 year record of discussing football still makes Jeff down the pub an authority – even if you only use his recommendations to precisely implement the opposite for your accumulator bet. 

Similarly, length of service is a good indicator of an individual’s capacity to speak authoritatively on a subject. While not all platforms remain forever, if you’ve previously made in-depth academic arguments on your domain specialism on MySpace (it’s still there, apparently, though we had to check, and we’d be very surprised if most of it worked anymore) and are worried about losing it, consider transferring it to your brand new LiveJournal profile (launched in 1999, yeah, we’re old – but we know everyone has a Substack now) with a ‘first published’ date.

Job title

It’s not a perfect indicator of authority – but it does generally coincide with length of service, so while a new graduate and a Head of Department may have the same opinion, it would be easier for an algorithm to infer that ‘authoritativeness’ resides with the latter over the former.

Trustworthiness

Trustworthiness is definitely an all of the above situation – but it’s also one of the more difficult to establish algorithmically, as trustworthiness is a very human perception and is far from a static or definable quality (especially as somebody one person considers trustworthy may be the exact antithesis of the term to another). Nevertheless, the Quality Evaluator Guidelines describe trustworthiness as follows:

Experience, Expertise, Authoritativeness and Trust (E-E-A-T) are all important considerations in PQ [page quality] rating. The most important member at the center of the E-E-A-T family is Trust.

Trust: Consider the extent to which the page is accurate, honest, safe, and reliable.

The type and amount of Trust needed depends on the page, for example:

• Online stores need secure online payment systems and reliable customer service.
• Product reviews should be honest and written to help others make informed purchasing decisions (rather than solely to sell the product).
• Informational pages on clear YMYL topics must be accurate to prevent harm to people and society.
• Social media posts on non-YMYL topics may not need a high level of Trust, such as when the purpose of the post is to entertain its audience and the content of the post does not risk causing harm.

Experience, Expertise and Authoritativeness are important concepts that can support your assessment of Trust[…]

Effectively, as the diagram suggests, properly indicating the EEA elements of EEAT contributes to the establishment of trustworthiness, which would be difficult to establish algorithmically and would likely lead to debate even if it were. 

Domain interaction and referencing

As mentioned in the expertise section – one way to algorithmically establish trustworthiness would be through referencing. Whether that’s in terms of the definition of a claim in schema, or through references to academic works and trusted sources during the course of compiling an argument, this is – arguably – one way an algorithm could determine the truthfulness of a source, and, therefore, its trustworthiness.

Proving it

The following are a number of techniques to help demonstrate various aspects of EEAT:

• Linguistic signals: integrate language into your content which signal personal interaction with a subject/object, including the overcoming of challenges, experiential and comparative language etc.
• External indicators: build links and citations (linked and unlinked) from highly relevant, trusted domains, ensuring they are descriptive (text around the mention supports inclusion).
• Community signals: earn citations from forums, social media, or other trusted sites where the subject is discussed.
• Formal accreditation: link to any online directory of accredited professionals in which you appear – both in your profiles and in schema using a [sameAs] field.
• Semantic and linguistic: use appropriate (not obfuscatory) domain-specific jargon and terminology, in service of clear, industry-specific communication.
• Domain referencing: in content, refer to other genuine domain experts or recent academic publications rather than popular discussions, demonstrating a deep interaction with the knowledge domain.
• Longitudinal data: establish and maintain a long-term record of your experience with and discussion of the subject to indicate length of service.
• Entity creation: make sure you’re consistently crafting your expert entity, that external profile information is well-reproduced across platforms, and is referenced by relevant third parties, and that it is obvious what your relationship is to the domain.
• Trustworthiness: this is generally built at an entity level by successfully demonstrating the EEA elements, but certain page types require specific trust signals. These include:
  • Secure systems: online stores require secure online payment systems and reliable customer service.
  • Content accuracy (YMYL): informational pages on clear YMYL topics need accurate content to prevent harm.
  • Honest reviews: provide honest product reviews written to help others make informed purchasing decisions, rather solely as promotion.

Final thoughts

We’ll end by repeating what we’ve tried to make clear throughout – most of this has been established to correlate with better performance, but we can’t say they’re causative, because Google won’t tell us. That being said, this is stuff we’ve been doing for a fair old while, and it’s worked so far (touch wood). 

Hopefully that shows at least experience, though whether we’re expert, authoritative and trustworthy is, we feel, for others to say. 

And thus ends our accidental trilogy of EEAT posts! We hope you find them useful, and that they can help you to improve your performance in search, even as things change and we move into a heavily AI-influenced era. If you’d like to speak about working with us – or about anything we’ve talked about over the last several weeks, you can contact us here.