By Derrick Edward
I've been thinking a lot lately about how I use the internet. Two years ago, if I needed to find a software solution or research a market trend, I'd hit Google, wade through ten blue links, dodge three ads, and hopefully find an answer.
Today? I just ask Perplexity or ChatGPT.
I'm convinced that we are witnessing a fundamental shift, not just in "search," but in how humanity consumes the web. AI isn't just replacing Google; it's replacing the act of web browsing itself.
But being a technical engineer, I wanted to answer the ultimate question (and the reason we built AuraScope). How do these models decide to recommend your brand over your competitor's?
I'm bullish that technical founders, outsiders to the traditional marketing world are actually best positioned to solve this. Why? Because Answer Engine Optimisation (AEO) isn't a keyword problem; it's an intelligence problem.
Traditional SEO was about understanding a list of links. AEO is about understanding a model. To win here, you have to understand the engineering of how that intelligence retrieves, processes, and decides on information. That is a very technical problem, and it's one we are uniquely positioned to solve.
I dove deep into the mechanics of this, validating our own engineering against industry research. And what I've discovered is this: The rules of SEO haven't just changed; the game board has been flipped. Here is what is actually happening inside the "Black Box," and what businesses in Aotearoa and Australia need to do to ensure they aren't invisible to the new gatekeepers.
The Two-Step Dance: Retrieval vs. Re-ranking
AI Search is not magic, it's a complex engineering workflow. When you ask ChatGPT a question about your industry, it doesn't just "hallucinate" an answer from nowhere. It follows a logic path that separates the winners from the invisible.
1. The Old School Retrieval
First, the AI uses traditional search methods (like keywords and indices) to grab a bucket of potential results. If you have good traditional SEO, you make it into this initial bucket. You're in the consideration set.
2. The Semantic Re-ranking (The "Aura" Filter)
This is where the magic and the danger happens. The AI takes those raw results and "re-ranks" them based on relevance before it generates an answer.
The AI judges your content differently than a human.
If your page title is clickbait, optimised to get a human to click, the AI looks at the actual content, realises the metadata doesn't match the substance, and filters you out. It will skip your high-ranking SEO page for a competitor's page that is boring, descriptive, and fact-dense.
The Insight: High "click-through" optimisation can actually hurt your "Share of Answers" in AI. You need to optimise for the machine's understanding, not just human curiosity.
Three Ways You Are Accidentally Ghosting the AI
We see this constantly when we run Health Scores for Kiwi businesses. You have a beautiful website, great traffic, but to ChatGPT, you are a ghost.
1. The JavaScript Trap
I cannot stress this enough, if you are relying on client-side JavaScript to render your content, you are playing a dangerous game.
Most AI search indexers (the bots that feed ChatGPT and Perplexity) simply do not execute JavaScript. They don't have the time or compute power to wait for your fancy React app to load.
If your content isn't rendered on the server side (SSR), the AI sees a blank page. It doesn't matter how good your product is; if the bot can't read it, you don't exist in the answer.
2. Fluff vs. Facts
If you're writing 2,000-word blog posts full of fluff to keep people on the page longer, AI hates that.
When users interact with AI, they are usually in "problem-solving mode". They want the answer, the code snippet, or the comparison table.
If your site is full of marketing "glaze", superlatives, vague promises, and sales fluff, the AI struggles to extract the facts. The brands winning in our Visibility Scores are the ones providing dense, structured, descriptive context about what their product actually does.
3. Conflicting Truths
This one is fascinating. We see companies with five different pricing pages floating around from 2021, 2023, and 2025.
When an AI like deep research digs into your site, it consumes more content than a standard search. If it finds contradictory information (like old pricing), it gets confused. It might hallucinate the wrong price, or worse, deem your source unreliable and cite someone else.
From Black Box to Glass House
It's easy to feel like this is all out of our control. But it's not. It's just data.
At AuraScope, we talk about moving from the "Black Box" (hoping AI likes you) to the "Glass House" (seeing exactly what it sees).
You don't need to guess if you're winning. You need to measure it.
- Are you being cited? (Citation Forensics)
- Is your technical setup blocking the bots? (Health Score)
- Is your content "fact-dense" enough for re-ranking? (Visibility Score)
The Opportunity for Aotearoa
I love that Kiwi and Aussie businesses are often forced to be more resourceful. We don't always have the massive ad budgets of the US giants.
But this shift to Answer Engine Optimisation (AEO) is a massive equaliser. It's a positive-sum game right now. If you create clear, helpful, technically accessible content, you help the AI, you help the user, and you win the citation.
It's not about tricking the algorithm anymore. It's about being the best answer. We're still early in this. If you're a marketing leader or founder feeling a bit lost in the AI noise, you're not alone.
We're all figuring out the new rules together. But I'm betting on the brands that turn the lights on.
If you're keen to see your own Brand Aura, we're opening up beta access to our Multi-Model Diagnostic tool. No sales pitch, just data. Reach out if you want to take a look.
