Google told marketers, in plain language, what it's rewarding in AI search right now. Most brands haven't read it. Here's the short version — and what to actually do about it.

Google said the quiet part out loud

In May, Google published an update on how AI Mode and AI Overviews decide what to show people. Buried in product-update language is a fairly direct statement of intent:

Google wants its AI answers to connect people with "authentic voices" and real firsthand experience, not just well-optimised pages.

‍

Five changes came with it. Most are minor UX tweaks. One is a big deal for brands.

Google is now surfacing previews of real people's opinions directly inside AI answers — pulled from public discussions, social platforms, and other firsthand sources, shown with the person's name, handle, or community attached. Search for advice on a topic, and instead of just a synthesised summary, you may now see an actual quote from an actual person, credited, with a link to where it came from.

Google didn't say exactly which platforms feed this. Given Google owns it and can structure it reliably, YouTube is the likely anchor source — but we’ve seen Instagram content in the wild being included as well so we can assume that social in general is becoming important.

Quick primer: what is E-E-A-T?

If the phrase "authentic, identifiable voices" sounds like it's pointing at something specific, it is. Google has a name for the quality bar it's describing, and it's worth knowing even at a glance: E-E-A-T, short for Experience, Expertise, Authoritativeness, and Trustworthiness.

It's the framework Google has used for years — originally to train the human reviewers who quality-check search results, and increasingly as shorthand for what its algorithms and AI systems are built to reward. In plain terms, each letter is really just a question:

● Experience — did the person actually use this, or are they just describing it secondhand?

● Expertise — do they genuinely know the subject, or are they guessing?

● Authoritativeness — is this person or source recognised as credible on this topic?

● Trustworthiness — is the source honest, transparent, and not hiding a conflict of interest?

You don't need to memorise the acronym. The useful takeaway is simpler: Google has always had a rough model of "what makes content worth trusting," and the May update is that model showing up more visibly inside AI answers — Google surfacing the specific people and sources it judges to score well on it, credited by name.

What this actually means for your brand

Strip away the product-update framing and the message to brands is this: Google's AI is now actively looking for real people talking about real experiences, and showing its work when it finds them.

That's a shift worth thinking about: For twenty years, "getting found" meant your own website ranking well. Increasingly, it means someone else's honest opinion about you being the thing Google chooses to surface — with your brand along for the ride only if that opinion exists somewhere Google can find it, and only if Google trusts it enough to show it.

A few jump out from this:

● Being well-optimised isn't enough on its own. Google explicitly said it's rewarding "authentic, identifiable voices." A perfectly written product page competes with a real customer's real opinion — and increasingly loses. AI engines are typically not quoting brands, but only third party sources.

● To be found by Google AI you don't need to rank #1 anymore. Google confirmed it's using a technique that goes looking beyond the top results to answer specific sub-questions people are really asking. A page that answers one specific question really well can get pulled into an AI answer even if it would never crack page one of classic search.

● Attribution is back. Google is now showing who said something, not just what was said. That means a credible, identifiable source is worth more than an anonymous one.

Simple checklist

You don't need a technical SEO team to act on this. You need to ask five questions about your brand's presence right now:

1. Does real, identifiable people's content about your products exist anywhere public? Reviews, videos, forum threads, social posts — not brand copy, other people's words.

2. Is that content clearly attributed to a real person or handle? Anonymous or vague sourcing is exactly what Google is now designing around, not for. Remember also that content on Reddit is anonymous.  

3. Does any of it answer one specific question really well? "Is this good for X" beats generic praise every time under Google's new retrieval approach.

4. Would you trust it if you were a stranger reading it cold? That's roughly the bar Google's AI is now trying to apply.

5. Is it easy to find, or is it buried in a comments section three scrolls down? Discoverable and well-organised beats vague and scattered.

If the honest answer to most of these is no, that's not a content problem you fix with better copywriting. It's an evidence problem — and it's the one your competitors are least likely to have solved yet, because almost nobody has read this update closely.

Going deeper

Getting found is the first problem. Getting confidently recommended — rather than hedged on — is the second, harder one, and it comes back to E-E-A-T: how AI decides it can trust what it finds well enough to stake a recommendation on it, not just how well the content is written.

We've written a deeper guide on exactly that — how E-E-A-T actually works, and where the current playbook for it falls short — for anyone who wants to go past "what to do" and into "why it works the way it does." Worth a read if this piece raised more questions than it answered.

‍