If you’re not visible in AI discovery today, you’re already invisible to the audiences that matter most. The brands that adapt now will own the conversations shaping tomorrow’s decisions.
Our data makes the shift unmistakable. AI referrals are occurring four times more often on desktop than mobile, while mobile organic traffic is declining at twice the rate of desktop. The first wave of adoption is coming from desktop-first knowledge workers and decision-makers, precisely the audiences that matter most to B2B brands, enterprise companies, and high-consideration consumer markets.
For businesses, the implication is urgent. Visibility is no longer about ranking highest. It’s about being cited, trusted, and remembered inside the generative engines already shaping these decisions.
Desktop-First Behaviour is Where GEO Begins
Why are AI referrals so much higher on desktop? Because desktop users are more likely to engage in complex, multi-turn discovery. Researchers, analysts, SEOs, and knowledge workers are crafting nuanced prompts, running multi-step conversations, and using AI systems as collaborators.
Mobile, by contrast, is built for quick answers.
With less screen space and more interruptions, mobile users often accept the AI summary at face value and move on. That explains why mobile organic traffic is dropping at twice the rate of desktop traffic. There’s simply less incentive to scroll, click, or explore beyond the first AI-generated answer.
This dynamic matters. Right now, AI discovery is shaping the journeys of the audiences that count most: B2B buyers, enterprise decision-makers, and executives signing off on budgets. These aren’t casual scrollers; they’re running research-heavy processes where trust and authority decide the shortlist. For Kiwi brands, this is the premium lead pool. Get cited in their discovery flow and you’re in the room where the deals are made. Miss it, and you’re invisible before the pitch even starts.
The Great Decoupling on Mobile
On mobile, especially, we’re seeing the “Great Decoupling” of visibility from traffic. Content still fuels the system, but the AI becomes the destination. Google’s AI Overviews, for example, are designed to fulfil the query without sending people elsewhere, and on smaller screens, traditional organic listings are pushed deep below the fold. For businesses, this decoupling means that being seen is no longer the same as being visited. Your content may shape decisions, but unless you’ve adapted for GEO, it won’t drive outcomes.
For Kiwi users, this is already everyday behaviour, with 80% of online shoppers now using mobile devices to browse and make purchases. Someone searching for a restaurant, comparing products, or looking up a quick how-to rarely scrolls past the AI Overview. On a phone screen, the synthesis feels like the final answer. That means your brand can be visible, even influential, without ever recording a click. Local visibility is happening, but without the visit.
Meet the New Gatekeepers
Discovery is no longer centralised. Google isn’t the only front door, and it’s no longer just a front door at all. AI platforms don’t simply retrieve content; they orchestrate intent. Each AI platform acts as a gatekeeper, each with its own architecture, biases, and opportunities. To win visibility, brands need to understand who controls the pathways and how to work with them.
For New Zealand businesses, this changes the game. It’s not about ranking in Google anymore; it’s about showing up wherever those intent fragments land: Google AI Overviews, ChatGPT, Perplexity, Bing Copilot, or Claude. Visibility now demands a multi-platform GEO strategy, because the answer your customer sees tomorrow might be assembled from five different engines today.
Google: AI Overviews & AI Mode
Google remains the most influential player, but the rules of engagement have changed.
- Dominant reach. AI Overviews, launched in 2024, now appear in more than half of all searches, making them the single most widely used generative product globally. By mid-2025, they had rolled out to more than 1.5 billion users.
- Conversational evolution. With AI Mode, launched in May 2025, search is no longer a static results page. It’s a full AI-first interface where users engage in multi-turn dialogues and exploratory queries.
- Extreme personalisation. No two people see the same output. Google pulls from Gmail, YouTube, Maps, Shopping, Calendar, and more, blending behavioural and contextual data to generate deeply individualised answers.
- Impact on clicks. Ahrefs analysed 300,000 keywords and found that when an AI Overview was present, clickthrough rates were lower by 34.5% compared to similar informational keywords without an AI Overview.
- The GEO play. To appear, content must be cleanly extractable, entity-rich, and semantically clear. Google’s “query fan-out” process breaks one prompt into dozens of subqueries, pulling snippets from across its index. Only content with strong E-E-A-T signals and machine-readable structure will survive that filtering.
Bottom line: In Google’s AI-first world, if your content isn’t structured for multi-intent retrieval, you won’t be cited.
OpenAI: ChatGPT
OpenAI has become the de facto conversational layer for discovery, a new kind of front door to information.
- Massive scale. ChatGPT processes an estimated 2.5 billion prompts daily, around 15–20% of Google’s daily search volume. It has 800 million weekly users, generating 0.11% of all global web traffic in 2025, nearly 4x higher than its 0.03% average in 2024.
- Multi-turn dominance. Over half of ChatGPT conversations are multi-turn, with the model carrying constraints and context forward automatically. It’s designed for dialogue, not one-off queries.
- Architecture. Unlike Google, ChatGPT doesn’t maintain a full index. Instead, it generates queries to Bing (and increasingly Google), fetches full URLs in real time, and synthesises results on the fly.
- The GEO play. Visibility here relies on real-time retrievability. Your content must be crawlable, fast, lightweight, and semantically transparent. Clear structure, authoritative tone, and robust metadata increase the odds of being pulled into an answer.
Bottom line: If your site can’t be retrieved and parsed in real time, you don’t exist in ChatGPT’s ecosystem.
Perplexity AI
Perplexity positions itself as the “transparent” alternative, and for GEO practitioners, that’s gold.
- Transparency-first. Unlike other platforms, Perplexity displays its citations prominently, often even before the generated text. This transparency gives brands direct visibility when they’re included.
- Rapid growth. In May 2025, Perplexity processed 780 million queries and attracted 129 million visits. It’s 15 million monthly active users make it small compared to Google, but its growth rate, 20% MoM, is explosive.
- Content preferences. Perplexity rewards direct answer formatting, entity prominence, and visible expertise signals like author bylines, citations, and credentials. Clarity wins over fluff.
- The GEO play. Think of Perplexity as a “truth table” for generative optimisation. Because it shows you exactly what it’s citing, you can reverse-engineer visibility signals and apply those lessons across less transparent platforms.
Bottom line: Perplexity is where brands can study what works and earn early credibility before larger platforms catch up.
Microsoft: Bing Copilot
Copilot isn’t just a search product; it’s baked into the daily workflow of millions of knowledge workers.
- Embedded everywhere. Integrated into Microsoft 365, Outlook, Excel, Word, Teams, Edge, and Windows 11, Copilot has access to more than 400M paid Microsoft 365 seats and 1.4B Windows devices. Even if monthly active users are lower (around 33M), the potential reach is staggering.
- The action layer. Unlike Google or ChatGPT, Copilot makes outputs instantly usable. A summary can be exported to Word, a data list to Excel, or an itinerary to Teams. For professionals, this is game-changing.
- The GEO play. Copilot loves structured, reusable content. Tables, lists, CSV-friendly formats, and well-structured HTML are disproportionately valuable here. Classic SEO hygiene, crawlability, canonical tags, and speed also pay off more than ever.
Bottom line: If your content is actionable, Copilot will surface it. If it isn’t, you’ll be invisible in the productivity layer.
Anthropic: Claude
Claude isn’t a consumer search product, but its influence in professional contexts is growing fast.
- Trust and safety first. Claude’s models are designed to be helpful, harmless, and honest. Accuracy and ethical framing are non-negotiable.
- Enterprise integrations. Claude is embedded in Slack, Poe, and enterprise platforms where fact-checking and reliability matter most. It has 18.9M monthly active users globally.
- The GEO play. Content must be factually consistent, ethically framed, and easy to synthesise. Strong sourcing, neutral tone, and alignment with professional standards are key.
Bottom line: For B2B and enterprise visibility, Claude is becoming a gatekeeper who rewards brands that can prove credibility.
Bottom line: Perplexity is where brands can study what works and earn early credibility before larger platforms catch up.
Winning in the GEO Landscape
Together, these platforms represent the new front doors to discovery. Google no longer has a monopoly on attention; discovery now flows through multiple AI intermediaries, each with its own logic. Some favour structured snippets, others favour transparency and citations, and others favour actionability inside productivity tools. The result: there is no longer one playbook; there are many.
For businesses, thriving in this environment means operationalising GEO as a discipline, a set of practices that go beyond classic SEO hygiene and adapt to the unique demands of each gatekeeper. Here’s what that looks like in practice:
1. Think in Systems, Not Keywords
Classic SEO rewarded keyword targeting for a single platform. GEO demands content that can survive in a fragmented landscape. That means engineering assets that are machine-readable, semantically clear, and resilient enough to be pulled, parsed, and reused across multiple engines. Instead of asking, “How do I rank for this query?”, the better question is, “How can this content be understood and reused across systems?”
2. Optimise for Extraction and Assembly
Every AI gatekeeper works by breaking content down into pieces and reassembling it into answers. That makes modular design essential.
- Tables and comparisons: Easy for AI to insert into Overviews or Copilot.
- FAQs and schemas: Ideal for quick retrieval and snippet generation.
- Charts, visuals, and PDFs: Valuable in multimodal outputs where text alone won’t suffice.
Content that’s easy to disassemble and reconfigure will get surfaced more often, regardless of platform.
3. Prioritise Trust as a Signal
In the generative era, trust is currency. Backlinks still matter, but AIs weigh authority differently: entity strength, clear sourcing, transparent credentials, and factual consistency across your ecosystem. If your brand doesn’t project credibility in a way machines can recognise, you’ll be passed over for one that does. E-E-A-T signals (experience, expertise, authoritativeness, trustworthiness) have moved from SEO “nice-to-have” to GEO essentials.
4. Prepare for Fragmentation
User journeys are no longer linear. A decision might start in Google, continue in ChatGPT, cross into Perplexity for verification, and finish in Bing Copilot as a ready-to-use Excel list. If your brand only shows up in one of those environments, you risk being cut out of the decision entirely. That’s why consistency matters. GEO success means building a presence that’s authoritative and recognisable across all gatekeepers, not just one.
The winners in the GEO era won’t be the brands that dominate a single platform. They’ll be the ones with a cohesive presence across all of them, cited, trusted, and machine-readable everywhere that matters.
Desktop Decision-Makers Are Leading the GEO Shift
The first wave of AI discovery is desktop-first, led by the knowledge workers and decision-makers who matter most for B2B and high-consideration markets. Miss them now, and you’ll be absent from the conversations shaping tomorrow’s contracts, purchases, and partnerships.
On mobile, the “Great Decoupling” is already biting; people skim the AI answer and stop. No click, no visit, no second chance. And the old Google-only playbook? Gone. Discovery is fragmented across Google, ChatGPT, Perplexity, Bing Copilot, and Claude, each pulling answers in its own way.
Trust and clarity are the signal in the noise. AI systems cling to brands that are structured, authoritative, and consistent, then reuse them again and again. Rankings might fade, but citations compound. The winners in the GEO landscape won’t be those shouting loudest; they’ll be the ones remembered, retrieved, and trusted every time the machines are asked to decide.
Read Chapter 4
