Google published something interesting last week: an official guide to optimising for AI Overviews. Not a blog post or a help document tucked away in Search Console, but a proper guide with technical explanations and specific recommendations. The fact they've done this at all tells you something important, even if what they're saying is more complicated than it first appears.
The guide itself is useful. It confirms several mechanisms we've covered in this series, particularly around retrieval-augmented generation and query fan-out. When someone asks Google's AI a question, the system generates multiple search queries behind the scenes, retrieves relevant content, then synthesises an answer. Google's now saying this explicitly, which validates the approach of optimising for the questions your content answers rather than just the keywords it contains.
They’ve also done the industry a favour by debunking some of the noise that’s been circulating. Those llms.txt files that were supposed to help AI systems understand your content? Google has confirmed they're unnecessary. Over-engineered structured data that tries to anticipate every possible AI query? Missing the point entirely. Generic, thin content created specifically for AI? Still fails, just like it always has for traditional search.
Here's where it gets interesting. Google keeps insisting this is "still SEO" whilst simultaneously publishing a dedicated guide that wouldn't need to exist if that were entirely true. The act of creating this document contradicts the framing.
If optimising for AI search were genuinely the same as standard SEO, there would be no need for a dedicated guide. The SEO Starter Guide already exists and has existed for years. Publishing something new, sitting alongside it in the fundamentals section, implicitly acknowledges that something has shifted enough to warrant separate treatment.
That's not to say Google is wrong that the foundations overlap. They do. But the act of creating this guide is more honest than the framing around it. You don't write a new playbook for a game that hasn't changed.
The strategic omission is more revealing than what’s included. The guide focuses exclusively on Google’s systems. There’s no mention of ChatGPT, Perplexity, Claude, Gemini, or any of the other AI platforms where your potential customers might be searching.
This isn't an oversight. Google's guide tells you how to optimise for Google's AI, which is valuable but incomplete if you're trying to build a visibility strategy that works across the platforms people actually use.
Search is fragmenting. People are finding information through ChatGPT, Perplexity, Claude, Gemini, and other AI platforms that have nothing to do with Google Search. AI optimisation is already a recognised discipline. Google publishing this guide confirms it. The conversation is expanding well beyond Google’s ecosystem. This isn’t a future trend to prepare for. It’s a current shift to act on.
That’s not a criticism. It’s just the nature of how any platform frames its own space. But Google’s guide is authoritative for Google’s systems only, and your customers are using more than one.
Don’t be misled by the 1% referral traffic figure you’ll see quoted for AI platforms. That number masks where the real activity is happening. 73% of B2B buyers now use AI tools during their purchasing process, and we’re increasingly hearing from clients that prospects mention having “looked them up on ChatGPT” before making contact. Tech industry professionals are using these platforms daily. Younger audiences are adopting them faster than any other demographic. The pattern is familiar: new technology gets adopted first at the edges, then it spreads. The businesses building their AI visibility now are the ones who will be hardest to displace when it does.
The pattern we observe most consistently is around entity clarity. Businesses that have invested in making sure their name, category, and description are accurate and consistent across the web tend to show up more reliably in ChatGPT, Perplexity, and Claude than in Google’s AI Overviews, where traditional domain authority still carries more weight.
Google's systems have years of ranking signal (data) to draw on. The newer platforms are leaning harder on what's clearly and consistently stated about you across multiple sources.
When NZ businesses actually run those searches, the most common finding isn't that they're missing entirely. It's that they're present but vague. The AI knows they exist but can't say anything specific or confident about them.
You get a generic description that could apply to any business in their category, no clear sense of what makes them different, and often a stronger mention of a competitor who has more independent content written about them across the web.
The second most common issue is category confusion. A business gets described primarily by the thing it's least known for, simply because that's what the loudest signal across the web happens to say. Both are description problems, not findability problems.
The business is findable, but what AI says when it finds them isn't doing any selling. That's actually the more urgent issue for most NZ businesses. Not "are we showing up" but "what is AI saying about us when we do."
Read it, but don't stop there. The guide is genuinely useful for understanding how Google's AI systems work and what they reward. If you haven't already addressed the fundamentals it covers, that's where to start. Make sure your content is crawlable, make sure it's genuinely useful rather than generic, and make sure your Google Business Profile is accurate and complete.
Those things matter, and the guide explains why clearly.
But then open ChatGPT, Perplexity, or Claude and search for your own business the way a prospect would. Not your brand name. Your category query. “Best commercial equipment suppliers in Auckland” or whatever the relevant version is for your business.
See what comes back. Are you in the answer? Are you described accurately? Are your competitors named and you're not?
Google's guide tells you how to be visible in Google's AI systems, and that's valuable. But it tells you nothing about what those other platforms currently say about you. For most NZ businesses, that's where the most surprising gaps are.
The guide is a good checklist for your Google presence. The category query test is how you find out what the full picture actually looks like across every platform your customers are using.
Source: Rod Russell, Managing Partner, ADMATIC, 18th May 2026