Richard Conway wants businesses to ask: when AI does more of the searching, comparing and acting, are we the brand it can understand, trust and use?

For most of my career, search marketing has been measured by visibility. Rankings, clicks, impressions, snippets, links, traffic. All these things are still relevant and are not going away however, our remit is expanding and the way our products and services are discovered online is changing fast.

Google I/O 2026 solidified what we have been saying for months and that a lot of Kiwi businesses are not up to speed about yet. The future of search is not only about whether a customer can find you. It is about whether an AI system can understand you, trust you, use you as evidence, recommend you in context, and increasingly act on the customer's behalf.

That is what grounding is, and it is the part most people are missing.

In Google's own technical language, Grounding with Google Search connects Gemini to live web content so it can give more accurate answers and cite verifiable sources beyond its training cutoff. In plain English it asks one question: can the machine find enough reliable evidence to believe you?

If an AI system cannot ground its answer in your brand, your expertise, your reviews, your data, your products, your pricing, your locations, your people and your proof, then you may not exist in the moment that matters. Not because your website is no longer optimised or valid, but because the machine cannot confidently use you. That problem is very different from traditional SEO and in our experience many Kiwi enterprises are not yet set up to manage this transition.

This was not just a product launch

I have spent more than 20 years in search, and I have watched a lot of "this changes everything" keynotes that have had minimal effect, I've even been caught up a few times by the hype! This one is different, because the numbers now support the shift.

At I/O 2026, Google confirmed AI Overviews now reaches over 2.5 billion monthly users. AI Mode has passed 1 billion monthly users in roughly a year, with queries more than doubling every quarter since launch. That is not a niche change, this is a mainstream challenge that is evolving daily.

Google also made AI Mode default to Gemini 3.5 Flash globally, introduced what it calls the biggest upgrade to the search box in 25 years (it now takes text, images, files, videos and even your open Chrome tabs), and made it seamless to move from an AI Overview into a conversation, with context carrying through your follow up questions. This context shift is really important to understand, a single query is no longer that, it's now part of a conversation or deep dive.

Traditional SEO was built around isolated queries. Someone typed "best mortgage rates NZ" or "SEO agency Auckland" and got a page of deterministic results. AI search does not work like that. People ask messy, multi part questions and get personalised, probabilistic results. They upload a screenshot, compare options, ask for trade-offs, keep going, then ask what to do next. Search is becoming less of a directory and more of a decision assistant. When delegated decisions become the norm, everything changes.

Google's guidance is right. It is also incomplete

Like with most enterprises, there are biases and motivations beyond what they say (like shareholder value). Foundational SEO still matters for AI Overviews and AI Mode, a website still needs to be crawlable, indexable, useful, technically sound and eligible to appear. There is no magic AI markup, no special file, no shortcut that suddenly drops you into generative results.

But "SEO still matters" is not the same as "nothing has changed." The foundations hold true; Technical health, helpful content, authority, page experience, structured information, brand trust are still critical. What has changed is the commercial question sitting on top of them and the way they are presented.

We used to ask about ranking organically in Google. In our experiments, being page 1 of Google does not mean you will be included in AI search. Now we need to consider if an AI system can confidently use us as a source when it forms an answer or makes a recommendation. That's the foundation of GEO, AEO and Agentic SEO, a practice that is evolving all the time.

GEO is about being visible and accurately represented in generated answers. AEO is about making your expertise answerable, extractable and useful. Agentic SEO or Agentic Readiness goes further again and asks whether your business is ready for AI systems that do not just answer, but research, compare, monitor, book, call and act. The real shift is in optimising for AI mediated decisions or delegated decisions.

Grounding is the new trust layer

In AI search, a brand needs to be groundable, being optimised for keywords is no longer enough. The system needs reliable context to work out who you are, what you do, where you operate, who you serve, why you are credible, and what third-party evidence backs your claims.

In New Zealand we have found a big gap between where businesses are now and where they need to be for this new era of discoverability. Many have human centric websites with little or no thought about how AI / machine readable they are. They rely on on-site testimonials and bragging, whereas now third-party recommendations and proof points are critical to the trust layer. The businesses that understand this now can build a moat which will increasingly benefit them as AI search becomes ubiquitous. We analyse many websites with vague language, thin case studies, weak author bios, inconsistent NAP details across the web, an incomplete or out of date Google Business Profile, and useful information buried deep in PDFs. Reviews live in one place, leadership content in another, product detail somewhere else again.

Whilst a human visitor may still be able to decipher things, AI systems are unlikely to, they may even piece things together incorrectly or present false information.

Grounding is more of an ambiguity problem than a content problem. As marketers it is our job to ensure that ambiguity is dealt with before AI agents access and present the information. Obviously your website still matters, but so do your reviews, your Google Business Profile, your LinkedIn presence, your video, your PR, your industry citations, your product feeds and the way other trusted sources describe you. The brands that win in AI search will not be the ones publishing the most content, they will be the ones building trust through demonstrating the clearest evidence.

Agentic search changes the commercial game

The part of I/O that should worry more boardrooms than it does is the move toward search agents. At the ElevenLabs launch in Sydney recently, I was chatting to the guy from Mastercard who spearheaded the first agentic purchases in New Zealand and Australia. Reinforcing this shift, Google announced information agents that run in the background, monitor the web and fresh data like finance, shopping and sports, and send back synthesised updates. It announced expanded agentic booking, and in select categories such as home repair, beauty and pet care, the ability for Google to call businesses on a customer's behalf. This is the future and where I think we will land in 2027 & 2028.

As with most things, the roll out will likely happen in larger markets first. The benefit of this is that we get a bit of a buffer to learn and iterate, what is in no doubt however is that eventually it will be coming to our shores and that companies who take it seriously now will reap the rewards.

Picture how a customer purchase behaviour changes, they may not search "best accounting firm for SMEs in Auckland" and click five sites. They may ask an assistant to shortlist three firms, compare them, check the reviews, confirm SME experience, read the pricing signals, find the proof, and suggest who to contact. In that journey you are not competing for a click, you are competing to be selected by an agentic system whose entire job is to reduce uncertainty.

The winners will build context, not just more content

The lazy reaction to AI search is to publish more and get lost in an ever-expanding sea of content. Smart marketers are shifting to building more useful context, structured around the questions people ask when they are deciding, not the ones that are easy to write.

Not "what is SEO," but "which SEO agency is best for an enterprise business in New Zealand," "what proof shows this agency understands GEO and AI search," "who has experience in regulated industries," "what does a good CRO programme look like for a bank," "can this provider work with our analytics partner," and "what are the risks and dependencies." That is where the future focussed marketers now spend their time and energy.

In practice it comes down to a few things:

• Clear entity information: ensure a machine knows who you are, what you do, where you operate and who you serve.
• Proof led content: real case studies, results, methodologies and expert commentary rather than assertions.
• Structured information: schema, FAQs, product and local data, author detail and service definitions.
• External validation: reviews, PR, citations, partnerships and trusted mentions.
• Freshness: information should be current and relevant. Stale proof is weak proof.
• Conversion: when AI sends you a customer who is ready to act, your landing experience has to finish the job.

Why SEO, GEO, AEO, CRO and data now belong together

I/O confirmed something that has been true for a while. Search is no longer just a channel. It is becoming an operating layer for customer decisions. Which means SEO cannot sit in a silo, GEO cannot be a side project, AEO cannot be a few FAQ blocks at the bottom of a page, CRO cannot wait until traffic shows up, and analytics cannot keep reporting last click conversions as if that tells the whole story.

They have to work as one system. SEO gives you the technical, content and authority base, GEO gets you into generated answers, AEO answers the questions buyers are really asking and CRO turns high intent visits into outcomes. Analytics shows where behaviour is shifting and where value is being created and Agentic SEO ties it together with the bigger question: when AI does more of the searching, comparing and acting, are we the brand it can understand, trust and use?

The takeaway

Google I/O showed what we have been saying for a while is coming to fruition. SEO is evolving and practitioners and marketers need to evolve with it.

The brands that treat AI search as a gimmick will go hunting for tricks, they will ask how to rank in ChatGPT or optimise for AI Mode as if there is one hidden hack. We've seen this space before, whilst paying to be in or creating a dodgy top 10 list may currently work, the real value isn't in these hacks, it's in creating reliable, machine-readable information, supported by the trust layer of third-party validation.

The smarter brands are building grounding systems. Clear expertise, strong proof, clean data, more visible authority, more useful content and more measurable journeys. A brand that stays consistent across every surface an AI might check, we call this the new SEO or Search Everywhere Optimisation.

Richard Conway is the chief executive of The Optimisers.