The most common question I get from founders right now is some version of: how do I start showing up in ChatGPT?

A few years ago, it was how do I rank number one on Google. The question has changed because the way people find things has changed. Instead of typing a few words into a search box and clicking through a page of links, people are having full conversations with AI, asking follow-up questions and taking its recommendations. Your next customer might have already read about you, compared you and made up their mind before they ever land on your website.

It's a lot to keep up with. The rules are being rewritten in real time, the acronyms keep multiplying, and half the advice out there contradicts the other half. So this week, I wanted to bring in two people who spend every day inside this problem.

Rushana Maksudova and David Andersen, both formerly at Google, help brands across Australia, New Zealand and the US get found in AI search. We talked through what has actually changed, what gets a brand recommended, where most companies are going wrong, and exactly where to start if you're starting from zero.

David: The mechanism is similar. People want information, and they search for it. For the last 25 years, that's meant Google or Bing. What's changed is twofold: the range of engines people use, and how they use them. People no longer ask a question, click a link and read the page. AI reads the pages for them and hands back a summary. The kinds of questions people ask have changed too, and that shift will keep going.

David: No. When people say SEO, they usually mean Google. But Gemini, ChatGPT, Claude and the rest each use different search indexes behind the scenes. Claude and Perplexity use Brave, which looks for very different things to what Google has rewarded for the last 20 years. So instead of one index, you're now trying to rank in four or five, and then get cited by the AI on top of that. Even within Google, the pages winning in AI Mode and AI Overviews often aren't the ones winning in regular search.

People also behave differently. The questions they ask ChatGPT aren't the ones they type into Google. A common pattern now is doing the research in AI, then a final Google search to confirm and click through. The foundations are the same (getting indexed, a sound technical setup, good content), but the strategy is quite different.

David: You should definitely be considering them together. Both of them matter. As I mentioned, a typical person's search journey involves both AI and search - that's a very common pattern.

But also, let's say you're setting up the technical foundations for a website, like the metadata. You need to make sure it works both for traditional search and for AI visibility. With the content you're creating, it's good to have pieces that rank well, earn links and show up in search, but you also want to think about how to optimise for AI. So a content strategy might have some pieces aimed at ranking particularly well on Google, and other pieces aimed at the different kinds of questions people ask through AI. Doing both at once generally helps. In our experience, if you optimise very hard for one of them, there'll be side effects on the other. It's not complete overlap, but there is overlap in the strategy between the two.

David: Yes, because each model has its own preferences. ChatGPT has recently started favouring website pages, like product and feature pages, over blog posts. Claude and Perplexity lean much more on individual blog posts. So a software development tool should probably target Claude heavily, and maybe ChatGPT. A consumer goods brand wants ChatGPT. It depends on your customers, their profile and even the country you operate in.

David: AI is looking for a direct answer to the question the user asked. So you want content that answers the full range of questions your buyer is likely to have. Not just bottom-of-funnel ones like “what's the best coffee in Brisbane”, but the questions they have right through the journey. And that content should be quotable, with sections the AI can easily lift out and repeat back. There are three layers to showing up:

So being in the search index isn't enough. You need to be picked as a relevant source, then quoted in the answer.

Rushana: With your customer, not with keywords. Before anything gets written, you need to know their pain points, who you're really targeting and what makes you different. A brand told me recently about a customer who said, “I asked AI, gave it a bit more context about my situation, and it came back with your name. That's how I found you.” People are having far more nuanced conversations now. So you need to understand the buying context of each customer persona, and you usually have several. What problems do they have? What's their situation? What triggers them to start looking? Build your content around that.

I think of AI as your sales rep. Customers have full back-and-forth conversations with it, and they can come in at any stage, from “I've got this problem” to comparing vendors right at the end. You want to give that rep as much accurate, useful information as you can, so it can sell for you and handle the objections for you. That means your features, pricing, policies and positioning all need to be publicly available on your site. Buyers ask detailed questions. If the AI comes back with “I can't confirm they handle this”, or “the competitor can, but these guys can't”, you've lost them. And they did all that research without ever talking to you.

It's also why your blog matters more than ever, but the posts that work now are about problems and guidance, not just keywords. Most businesses show up reasonably well near the end of the journey, because that's where their product information sits. The goal is to show up as early as possible. The earlier you're mentioned, the more likely you become one of the three options that person digs into.

David: They all matter, and they work together. To an AI model, it doesn't really matter where the content lives. The one caveat is claims you make about your own company on your own website, which AI is increasingly sceptical of.

Being featured somewhere reputable, like Fast Company, is really good external proof. It shows up for searches your website just won't. In general, having a wider footprint across the web beats almost anything else.

The old goal of getting onto other people's sites was backlinks, but that's becoming less important. What matters now is being part of the conversation. It's more about the mention than the link. Other blogs mentioning you, even paid content on other sites, is valuable in its own right, whether or not they link back. Once you're mentioned and you're in the user's mind, they start asking the AI more questions about you.

David: Two places. The first is llms.txt. Companies create the file, fill it with stuff and call it done. The reality is that almost every model ignores it right now. It's only used when an agent is exploring your site and wants extra context. It feels productive, and it's something to point to, but that's about it. The second is mass-producing AI content: top-10 lists, programmatic pages, hundreds of pieces at once. At that scale, the indexes keep punishing it. It works for a little while, then falls off fast.

Rushana: More isn't the answer. Prioritise quality and be surgical. You still need that foundational content, but it has to be well structured and answer the right questions. Not “let's spam a thousand pages and hope for the best.”

David: AI-written content isn't the problem in itself. Google has said it doesn't mind content that's partly or fully AI-written. What it cares about is whether the information is valuable, useful and unique. Funnily enough, AI models actually prefer quoting content that reads like AI wrote it.

The problem is how it's used. Giving AI your brand and asking for a hundred pieces covering every question is far less effective, and far less useful to readers. What works best for us is starting with real material: sales calls, customer conversations, case studies, real quotes and testimonials. Then use AI to repackage them with careful structure, and give it a final human pass. It's also great at turning documentation into guides, or long technical detail into a clear set of features, which is exactly the kind of content that surfaces when people ask AI about you. The more care you put into each piece, the more you get out of it.

Rushana: There's less of it, but it's better. Fewer people are actually coming to the website, but the ones who do are much more likely to convert. The rough stat is around four times the conversion rate for people coming from AI versus Google search.Because they've already had that back-and-forth conversation and worked through their objections.

Rushana: Converting them. Make sure the content that's there converts well, and have some really good funnel pages. One of the strongest signs you're getting AI traffic is a rise in branded search. Someone has a conversation with AI, puts your name into Google, lands on your homepage and checks you out. So strong product and conversion pages are more important than ever.

There's still a balancing act between what's good for people and what's quotable for AI, just like there was with keywords in SEO. The key is working out the most important things you need to say and putting them in clear, answer-style sections. The biggest difference is language. Vague, fluffy “we're the best” marketing copy is much less effective than “we have these features, and we cover this”. Specifics work far better with AI than claims that just sound impressive.

David: I think of it in two stages. Stage one is showing up at all. Stage two is controlling how you're mentioned and the story being told about you. For a new brand, it's all about stage one.

A new brand isn't going to win the big, broad queries, and it shouldn't try. It has a wedge: a competitor it's taking business from, a specific customer it's going after. You're not a shopping centre that appeals to everyone. With that in mind, I'd break it into five steps.

David: For them, it's more about stage two. The big one is finding where you're being misrepresented. Is everything on your site accurate and consistent? Even companies with SEO plans have gaps, and buyer's guides and proper product pages are missing from a lot of big sites. But more of the work is answering the questions and caveats floating around about you, and responding to competitors' claims. AI will often say a company doesn't have a feature, or quote the wrong price. Taking control of that story matters.

On top of that, create content at the problem-recognition stage. Write guides that solve your customers' pain points, and think of them as lead magnets you drop into AI. Before someone has even met you, you've helped them. You become a useful reference in the category, and that builds credibility.

So, back to the question founders keep asking me: how do I start showing up in ChatGPT? The honest answer is that there's no trick. It's knowing your customer, answering the questions they're actually asking, and being specific about what you do. The engines are already reading. Start giving them something worth quoting.

David: SEO is increasingly becoming AI search, and Google is heading that way. It's doing everything it can so people don't need to click through to websites to get information.

Over the last year, we've gone from “AI tells me what to do” to AI doing things for the user. With agents, payments and people getting Claude to act on websites, I expect visits to keep falling. Businesses will want to let people buy, log in or get things done without ever visiting the site.

Rushana Maksudova and David Andersen are the team behind BrandViz.AI, a platform that helps brands show up in AI search. It finds the technical and content gaps holding a brand back in ChatGPT, Claude, Perplexity and Gemini, drafts the fixes, and publishes them straight to your website. They work with companies across Australia, New Zealand and the US, from early-stage startups to global enterprises.