I Optimised My Site for AI Search (Before Most People Were Even Asking About It)

Everyone's still arguing about whether "AI search" is a real thing businesses should care about yet. I figured the only way to actually know was to test it on myself.

So alongside the usual SEO work, I started checking a different question entirely: not "where do I rank on Google," but "does ChatGPT, Gemini, or Google's AI Overview actually mention my site when someone asks a relevant question."

Turns out it already does. And it's growing fast.

Figure out whether optimising for AI search is a real, measurable channel today, or just a theory people are getting excited about too early — and if it's real, work out what actually earns a mention inside an AI-generated answer, not just a ranking position.

Ranking tools tell you your Google position. They don't tell you what ChatGPT or Gemini think your page is about. I started checking that directly — and found some genuinely surprising gaps. For one of my own key search terms, a big chunk of what AI engines associated with it turned out to be completely irrelevant job-market content, not "hire someone" content. Fixing that meant adding specific, concrete details AI models could actually latch onto — real technologies I use, a genuine accessibility commitment — not just more keywords.

The first time I checked, I found a single AI Overview citing my homepage directly — a nice one-off. But I wanted to know if that was a fluke or a trend, so I pulled three months of real Search Console data specifically for AI Overview appearances.

It wasn't a fluke.

AI OVERVIEW IMPRESSIONS — JUNE THROUGH SEPTEMBER, STRAIGHT FROM SEARCH CONSOLE

Daily impressions in Google's AI Overviews went from around 26 a day in June to over 160 a day by September — and it wasn't one lucky page carrying the whole number. 79 different pages picked up visibility, including brand new content published during the same three months, which told me this wasn't a one-off — it's a repeatable result.

Around the same time, I was weighing whether to lean my own positioning toward "freelance" specifically, or keep it more generic. Rather than guess, I tracked how AI models actually talk about me under each framing.

The gap was bigger than I expected. Averaged across ChatGPT, Gemini, and Perplexity, "freelance web designer sydney" gets me an average position of 1.7 and a 55% mention rate. Drop "freelance" and just search "web designer sydney," and it falls to position 4.0 with a 20% mention rate — less than half as visible. ChatGPT specifically goes from a 1.6 average position down to 3.7 without "freelance" in the query. Perplexity's gap is even starker: 1.9 versus 7.5.

One honest wrinkle: Google's own AI Overviews actually run the opposite way — it favours the plain "web designer sydney" phrasing slightly over "freelance." So it's not a universal rule, it's model-specific. But across the conversational AI assistants people are increasingly using instead of Google altogether, leading with "freelance" isn't just better for how a human reads the page — it's measurably better for how AI models talk about me too.