E-E-A-T beyond your blog: build authority for Google and AI search

TL;DR summary

• E-E-A-T is not a score and not a direct ranking factor. It is a term from Google's quality rater guidelines, and it describes the kind of results we can expect to see on a search results page.
• Content you publish yourself is a claim. Content someone else publishes about you is proof. Generative AI answers weigh proof, because the model has to decide on its own who to trust.
• In our market, authority is built by specialised industry portals.
• There are no shortcuts. Consistency and quality are what count.
• External mentions have a limited effect if your site has no author pages and no structured data.
• Measure it by checking every month what AI tools say about you, and by recording the source of every enquiry in your CRM system.

Eighty articles and still invisible

A marketing manager at a mid-sized company opens ChatGPT and asks: "Which companies sell lawn mowers for a garden of 1,000 square metres?" He gets five suppliers. His own company is not among them, although it has worked in this niche for a decade and has eighty articles on its website about exactly this topic.

This is not a fault in the model. It is a correct answer to the question the model is actually solving: who is recognised on the web as a relevant source for this field. Eighty articles on your own domain, with no confirmation from outside, are sometimes not enough.

Why the rule has changed

A classic search engine returned a list of results. You typed in keywords, you got ten links, and you decided for yourself which result to trust. The judgement about credibility was entirely yours.

A generative answer does not leave that judgement to you. The model has to choose who to cite, and for that it needs some measure of trust in the source. It has no access to your references, your business results or how satisfied your clients are, so it relies on the only thing it can measure: how often, and in what context, your name appears at other people's addresses.

The same shift is happening at Google, which now answers more and more queries with an answer instead of a list. The consequence is uncomfortable for anyone who built a content strategy only on their own blog. Content you publish yourself is a claim. Content someone else publishes about you is proof. The systems that assemble answers weigh proof.

In June 2026 the Munich Regional Court ruled provisionally that Google is legally liable for false or defamatory statements produced by its AI Overviews feature. The court separated AI summaries from traditional links in search results and held that text generated by AI is Google's own independent content, not merely third party data.

What E-E-A-T is, and what it is not

The abbreviation stands for Experience, Expertise, Authoritativeness and Trustworthiness. Google uses it in the guidelines for its quality raters, the people who rate search results by hand and so help calibrate the algorithm.

There is an important difference here that many agencies miss. E-E-A-T is not a switch inside the algorithm, and it is not a score stored somewhere on your site. Google has explained several times that it is not a direct ranking factor. It is a description of the kind of results the algorithm is trying to show.

So why talk about it at all? Because the signals raters use to judge credibility are the same signals language models process. Who is the author? Do they exist outside this one website? Who refers to them? Is the company recognised in its industry? That overlap is why work on E-E-A-T pays off twice today. In classic search results, and in the summaries generated by AI.

Which external websites actually count

Slovenia, where we do most of our work, is a small market with a limited number of quality portals where you can appear. What does pay off is appearing on portals with a long history behind them and a clear specialisation. Examples from our market:

• Nepremičnine Si21, a property portal
• Avto.info, a motoring portal
• Si21, a general news portal

All three have good authority and more than 30 years of presence online. A mention on any of them earns you the extra points that signal to search engines and to AI tools that you are an expert in that field. Every market has its own equivalents, and the task is to find them.

It also matters that you appear in other places:

• business media,
• industry associations,
• social networks,
• live events.

What does not work

Buying external links in bulk. A hundred links from directories with no content add nothing, because these are places the algorithms recognised as low quality a long time ago.

• Guest articles written only so that they exist. If the text is generic and any company could have signed it, it does not help anyone understand who you are.
• Publishing in campaigns. Three articles in one month and then nothing for six months build nothing. Systems that assess sources take consistency over time into account.
• Relying on volume. The belief that more content brings more recognition does not hold, and it does not hold here either.

The technical part most companies miss

External mentions have a limited effect if your own site does not link its content together properly. This is a common mistake among the people who manage websites.

If articles are signed by "the editorial team" or "administrator", the system has nobody to connect to external profiles. Every author needs their own page, with a name, a role, a field of work and links to the profiles where they are present.

These links also have to be written in a structured way, with the right mark-up in the code. Mark-up for the organisation and for the person, and the field that lists the same person's profiles at other addresses, are what tell the system that this is one and the same human being. Without it, your LinkedIn profile, your website and a mention in a magazine are three completely separate things.

The same applies to consistency of name, address and telephone number, everywhere you appear.

Finally, the page that external sources point to has to answer the question quickly. A visitor who arrives from an AI answer comes with a far more specific intention than a visitor from classic search.

On the websites we build on CMS Bananadmin, our own content management system, author pages, mark-up for the organisation and the person, and the linking of profiles are part of standard website development, not an extra you have to order separately. This is also why we treat the technical side as part of our SEO Optimization Services and not as a service of its own.

How do you know it is working?

Measuring whether your website appears in AI results is difficult, because the standard tools do not show it directly. There are still four approaches.

1. Check regularly what generative systems say about you. Ask the same questions every month and save the answers.
2. Watch where your traffic comes from. The share of traffic from generative sources is still small, but its structure is different.
3. Record the source in your CRM system. With every enquiry, note how the client found you. If your website and your CRM system are connected, this happens automatically, and after a few months you can see which activities actually bring in business.
4. Google has added a new report in Google Search Console that shows how many times your website appeared in AI summaries.

What you can do this week

Find the external pages where your company appears. Separate the ones that say something about what you do from the ones that are only an entry in a list. That first number is your starting point. Check as well that external links lead to the final address, with no redirects. This includes the redirect from the http to the https form of the address.

Open the articles on your website and check whether an author is listed. If there is no signature, or if the signature is generic, you have your first task.

Choose two external websites where it makes sense for you to appear, and pick an expert in your company who will write one article a month for them over the next six months. Two good external websites are enough.

Define your KPIs, so that you can check how you are doing.

At Kabi we review your current presence outside your own domain and show you what AI tools know about your company today. If the problem turns out to be in the technical set-up of the site, we solve it with a rebuild. If the problem is that nobody writes about you, talk to us about a content strategy that changes this.