Chapter 5: The Playbook – How to Win in AI Search

February 2025 was the month the ground shifted under SEO’s feet. For 20 years, the game was predictable: rankings drove clicks, clicks drove revenue, and the chain was measurable end to end. But that equation? It’s snapped clean in two.

What followed was a digital earthquake. Our data shows organic traffic down 10% year-on-year and search share slipping by the same margin. At the same time, AI referrals exploded 10x in just one year but still account for less than 1% of sessions. Worse, the revenue per AI-driven session is far lower than organic. Visibility is happening, but the value isn’t flowing through.

AI isn’t some shiny add-on feature. It’s already deciding who gets cited, who gets erased, and who makes it into the answers that shape real decisions. If your content isn’t built for retrieval, you’re invisible.

The good news? There is a playbook. One that stabilises the organic traffic you’re losing, while building early-mover advantage on the platforms that will dominate discovery tomorrow.

This is the AI Visibility Playbook. And it comes down to five moves:

1. Audit: See your content the way machines see it.
2. Restructure: Break it down into chunks that engines can grab.
3. Secure Authority: Build the trust signals AI can’t ignore.
4. Monitor Visibility: Stop flying blind; track, simulate, and adapt.
5. Reframe Teams: Upgrade SEO teams into AI Visibility Teams.

Let’s break it down.

Every winning playbook starts with one uncomfortable question: Where do we stand today? In the AI-first era, that question isn’t about keyword rankings or backlink profiles. It’s about how machines see you. Before you can optimise, restructure, or scale, you need to understand how your existing content and infrastructure are being interpreted or ignored by generative systems.

Think of this audit as a reality check. It’s about stripping away assumptions and seeing your site the way generative engines do: as chunks, entities, and signals. Without this first step, everything else is guesswork, and in AI search, guesswork is fatal.

Technical Accessibility

AI engines can’t use what they can’t reach. Robots.txt still matters; block the wrong agents, and you block your way into AI synthesis pipelines. If your site is slow to load, hidden behind heavy client-side rendering, or structured in a way that makes parsing difficult, you’re effectively invisible. Sitemaps (both XML and HTML) should be tidy, up to date, and inclusive.

Think of it this way: traditional SEO asked, “Can Google crawl and index my pages?” GEO asks, “Can multiple AI engines extract and reuse my passages at scale?” If the answer’s no, you’re out of the game before it starts.

Content Readability & Extractability

AI doesn’t read articles the way humans do. It slices them into chunks, often paragraphs, and evaluates each one in isolation. That means a 1,000-word article with a single gem buried halfway through will never surface. Every paragraph needs to stand on its own.

An audit here should ask:

• Does each passage have a standalone value?
• Are stats presented clearly and verifiably?
• Can this section be lifted cleanly into an answer without losing meaning?

Authority & Trust

E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) isn’t optional anymore. AI systems don’t want generic pages; they want credible signals. Your audit should include a trust check: are your authors named and credentialed? Do you cite reputable sources? Do your case studies and data points provide information gain, something beyond what’s already on the web?

Why the Audit Matters

An audit is the compass that guides every GEO move that follows. It tells you where you’re already GEO-ready, where you’re invisible, and where your authority is falling short. Most importantly, it translates directly into action: fix accessibility before chasing citations, restructure unreadable blocks before publishing new content, and strengthen E-E-A-T before expecting trust. The payoff is clarity. Instead of optimising blindly, you’re making decisions grounded in how machines actually perceive your brand. In a generative world, that clarity is leverage, the foundation that turns visibility into influence.

Once you know where you stand, the next step is restructuring your content. This restructuring doesn’t mean scrapping everything you’ve written and starting again. Instead, it’s about reshaping what you already have so it’s machine-friendly. In the AI-first era, your content isn’t competing for a human’s attention first; it’s competing for a machine’s ability to parse, chunk, and reassemble it into an answer. If the machine can’t understand your content quickly, you won’t be surfaced, no matter how brilliant your message might be.

Semantic Units & Chunking

Big paragraphs are content graveyards. AI systems like Gemini and ChatGPT don’t read top to bottom; they chop pages into chunks and rank each one separately. That means your golden insight buried halfway down a 600-word section will never see daylight.

The fix: one idea per paragraph. Make it short, make it clear, and make sure it can stand on its own. Use headings and subheadings as signposts. Turn long explanations into bulleted steps or comparison tables. Every chunk of content should be “quote-ready” for AI.

Semantic Triples

Machines don’t think in long-winded explanations. They think in relationships: subject → predicate → object.

• “Xero integrates with over 1,000 apps.”
• “A heat pump reduces home energy use by 30%.”
• “Perplexity cites sources before providing answers.”

These crisp, factual statements are fuel for generative engines. They’re quick to map, easy to validate, and simple to slot straight into an answer. Work them into your content and you’re effectively hard-wiring your expertise directly into the machine’s knowledge graph.

Clear Entities

In AI search, entities matter more than keywords. Engines resolve meaning by mapping entities into their knowledge graphs. If you say “this tool,” you’ve lost. If you say “Google Search Console,” you’re in.

Use precise names. Keep terminology consistent. Link entities back to authoritative sources where possible (Wikidata, schema markup, internal pages). You’re training AI to understand exactly who you are, what you do, and how you connect to the bigger picture. Get sloppy and you’ll disappear.

Structured Data Beyond Schema.org

Schema markup is the starting line, not the finish. Yes, use FAQ Page, How To, Product, and Organisation schema. But if you stop there, you’re leaving equity on the table. Leaders are building custom ontologies and internal knowledge graphs, essentially turning their websites into structured databases.

This scaffolding helps AI slot your content neatly into its knowledge ecosystem. Without it, you’re just another messy page. With it, you become a trusted node in the network.

Multimodal Parity

AI search isn’t text-only. It pulls from images, videos, PDFs, tables, and transcripts. If your content only exists as a wall of text, you’ve limited your surface area. Repurpose it: turn guides into comparison charts, record explainer videos with transcripts, and add alt text to images. Every format is another entry point for retrieval.

Restructure or Disappear

Restructuring is about removing friction. Don’t make AI work to figure out what you mean. If your content is chunked cleanly, written with semantic triples, anchored in clear entities, supported with structured data, and expressed across formats, then engines can grab it instantly. And if they can grab it, they can cite it.

The generative web is drowning in noise. Most of it is recycled, shallow, or outright synthetic. Engines know this, and they’re scrambling to separate the credible from the questionable. That’s where authority comes in. If you don’t signal trust at a machine level, you’ll never surface for the human who matters.

E-E-A-T as the Price of Entry

Think of E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) as your ticket to the game. Without it, you’re not even on the field. Author bios, credentials, and transparent sourcing are structural signals that machines use to decide whether to cite you.

Experience is especially valuable because it’s something AI cannot fabricate. A case study showing how you solved a problem, a real-world experiment with messy results, or a post-mortem on what didn’t work, these are authority gold. They prove there’s a human with lived expertise behind the words.

Proprietary Data & Unique Insights

Generative models love originality because it reduces the risk of hallucination. If your content brings something new, it becomes far more “quotable.” A bland statement like “most businesses struggle with adoption” is forgettable. But “67% of New Zealand SMEs adopted AI tools in 2025” is specific, verifiable, and valuable. Proprietary data is one of the fastest ways to cut through the clutter and signal that your brand isn’t just another summariser, it’s a source.

Co-Citation & Corroboration

Authority doesn’t exist in a vacuum. AI engines are probabilistic: they lean on corroboration. If multiple trusted sources echo the same fact, it survives the synthesis process. Better yet, when your content is consistently cited alongside respected brands, you get pulled into what we call “credibility clusters.” In practice, that means your brand gets treated as part of the trusted network. It’s why contributing data to industry roundups or getting your name into analyst reports pays dividends.

Authenticity Matters Too

Authority isn’t always about polish. For queries where people want lived experience, troubleshooting, product reviews, and hands-on advice, AI often surfaces authentic voices. That might mean a Reddit thread, a how-to comment, or a customer review with gritty detail. Community-driven signals matter because they feel human. Sometimes authority is less about sounding corporate and more about sounding real.

Trust Is the Gatekeeper

Authority is the filter through which AI decides who gets amplified and who gets erased. Build it deliberately, make it visible, and you’ll earn a seat at the table where decisions are being made.

AI search is a black box. You can’t see the impressions, you can’t see the citations, and you certainly can’t log into a neat dashboard the way you do with Google Search Console. That doesn’t mean you get to shrug and hope for the best. If you want to compete in the generative era, you need to build your own visibility layer. Otherwise, you’re flying blind while your competitors engineer their way into the answers.

The Three Layers of Measurement

To make sense of AI visibility, think in three layers: input, channel, and performance.

• Input metrics are about eligibility. Are your passages retrievable? Do AI crawlers like PerplexityBot or ChatGPT-User actually hit your site? Do your paragraphs match the semantic shape of fan-out queries? These are early warning signals. If you’re not eligible at this stage, you’ll never be cited.
• Channel metrics are about presence. Is your brand actually showing up inside AI surfaces like Google’s AI Overviews or ChatGPT’s real-time responses? How often? In what position? Being cited first in a synthesis is the new equivalent of ranking #1. You need scripts and monitoring tools to query engines, capture the output, and measure your share of voice.
• Performance metrics are about impact. Are AI citations moving the needle? That means segmenting analytics to see if AI-affected landing pages are still converting, tracking whether branded search volume grows when your brand is mentioned in AI outputs, and measuring indirect lift even if clicks are low.

Each layer tells part of the story. Together, they form a visibility stack you can actually act on.

Build Your Own Instruments

There’s no “AI Visibility Console.” If you want the signals, you have to build the instruments. That means log file analysis to spot AI bot activity. That means browser automation frameworks like Puppeteer or Playwright to query AI systems at scale, capture the generative output, and parse the citations. That means retrieval simulations using tools like LlamaIndex to see which of your passages get selected when a vector search is run.

Yes, it’s work. But it’s also your competitive moat. The businesses that invest in measurement infrastructure today will have visibility into the ecosystem while everyone else is guessing.

Embrace Probabilistic Visibility

AI search isn’t deterministic like Google rankings. The same prompt can produce different outputs at different times. That’s not a bug; it’s how generative systems work. So don’t expect a static ranking report. Instead, think in distributions. How often does your brand appear in over 100 tests? How consistently do you show up alongside competitors in a credibility cluster? These probabilistic patterns are the real KPIs of AI visibility.

Turning Guesswork into Strategy

Monitoring AI visibility is the only way to close the Measurement Chasm and prove your influence in an ecosystem where clicks no longer tell the whole story. The brands that track, simulate, and refine their footprint will stop flying blind and start steering the answers themselves. Everyone else will keep guessing while the machines quietly decide their fate.

GEO is a whole new operating system. And when the operating system changes, the team running it has to change too. The skill sets, the roles, even the way we measure success, all of it needs a reboot.

The days of “keyword hunters” and “link builders” driving strategy are done. AI engines don’t care about blue links or ten-pack SERPs; they care about semantic clarity, entity relationships, and trust signals they can actually parse. To compete, your team has to evolve into an AI Visibility Team, a crew built to engineer relevance inside generative systems.

Roles of the Future – Building NZ’s First AI Visibility Teams

Generative AI has changed not just how people search, but also how organisations need to be structured if they want to win inside this new ecosystem. Stanford’s 2025 AI Index Report shows 78% of global organisations are now running AI in production, up from just 55% in 2023. Job postings mentioning generative AI skills have exploded, with some categories seeing over 2,000% year-on-year growth.

That wave isn’t stopping at New Zealand’s border. Datacom’s 2025 State of AI Index shows 87% of NZ organisations now use AI, up from just 66% in 2024. These shifts aren’t theoretical; they’re already transforming job roles here.

So which roles matter most? Currently, three stand out as pioneering positions within GEO teams. Each represents a different piece of the puzzle in how AI engines discover, interpret, and cite your brand in answers.

Relevance Engineer – The System Builder

Relevance engineers are the new hybrid of technical SEO, content architect, and data engineer. Where yesterday’s SEOs tweaked pages, a Relevance Engineer designs content systems: semantic structures, schemas, and knowledge graphs that AI can reliably parse and reuse.

Think of it as moving from fixing cars to building engines. Their job is to engineer relevance across multiple queries, contexts, and platforms. The mission isn’t to rank; it’s to make your brand the answer.

Retrieval Analyst – The Black-Box Decoder

If the Relevance Engineer builds the system, the Retrieval Analyst tells you how well it’s working inside the machine. They live in the data: testing AI outputs, tracking citation frequency, reverse-engineering fan-out queries, and mapping why competitors get cited when you don’t.

They monitor new KPIs like chunk retrieval frequency and entity citation density, turning invisible AI behaviour into insights that teams can act on. For NZ brands competing in crowded global niches, this role is the difference between “occasionally showing up” and “owning the conversation.”

AI Strategist – The Navigator

The AI Strategist sets the course. They look across the whole AI ecosystem and decide where to play, how to show up, and what’s coming next.

They also do the hardest job: changing the mindset inside the business. They help leadership stop thinking about GEO as “just another SEO channel” and start treating it as a core visibility system. In a market like NZ, where budgets are tight and first-mover advantage matters more than brute force, this role ensures the organisation moves early, not late.

Essential Skills for the AI Era

The toolbox of an AI Visibility Team looks very different from an old-school SEO team. Key skills include:

• NLP Fundamentals: Understanding entity recognition, embeddings, and intent classification. If you don’t speak machine language, you can’t optimise for it.
• Scripting & Automation: Python for building scrapers, parsing logs, and running simulations. Manual audits won’t cut it when AI updates weekly.
• Vector Embeddings: Knowing how models calculate meaning in high-dimensional space, and how to engineer content that aligns with it.
• Content Strategy for Machines: Crafting “fraggles” (fragmented passages), semantic triples, and modular content ready for extraction and reuse.
• Prompt Engineering: Using prompts to probe, test, and reverse-engineer how AI retrieves and cites information.
• Data Science Basics: A/B testing, statistical modelling, and visualising patterns across thousands of generative outputs.
• Knowledge Graph Management: Structuring your brand’s entities so machines know exactly what you do and how you connect to the bigger picture.

The Cultural Shift That Decides Who Wins

Survival in the generative era won’t come from tinkering at the edges. It demands a complete reset in how teams think about visibility. Traditional SEO was about nudging the system: polish a meta description here, shuffle a few keywords there, chase some links, and wait for Google to reward you. That mindset won’t survive the generative era.

GEO demands something different. It’s about designing systems that can be parsed, reused, and trusted by machines across multiple platforms. It’s about engineering reliability into how your brand gets retrieved and cited, not crossing your fingers for a lucky ranking.

The hardest part of this shift won’t be technical. It’ll be unlearning habits built over 20 years of SEO muscle memory. The businesses that make that leap, that reframe visibility as a machine-driven system, not a human popularity contest, will be the ones who seize the first-mover advantage.

Don’t just survive the shift. Own it. Don’t just cling to the old playbook. Write the new one and make sure it’s yours.