Here's an uncomfortable truth: your customers have probably stopped scrolling past the first Google result. These days, they're just asking ChatGPT, Claude or Perplexity to do the comparison for them.
At our latest Lunch & Learn, Darren sat down with Lucia, founding member of technical staff at Inflect, to unpack what this shift actually means for founders and how to make sure your brand shows up when AI is doing the recommending. Catch the full session on our Youtube Channel.
The shift from search to AI answers
Traditional search starts with a keyword: "best shoes for flat feet." AI search starts with a situation: "I run three times a week, my feet hurt, I've got a $150 budget, what should I buy?"
That context changes everything. A business no longer just needs to rank for a keyword, it needs to be understood well enough to be recommended for a specific person's situation. That's the essence of GEO, or generative engine optimisation: helping AI systems understand, trust, cite and recommend your business.
Is GEO just SEO with a new name? Not quite. SEO gets a page discovered and ranked. GEO is about whether the brand actually gets mentioned and recommended inside the AI's answer.
Why ranking #1 on Google isn't enough anymore
One study Lucia shared analysed 300,000 keywords and looked at pages ranking first on Google. Even when those pages held their top position, click-through rate dropped by 58% once Google's AI overview appeared. People are reading the AI summary and never clicking through at all.
This doesn't mean SEO is dead. It means rankings and clicks no longer tell the whole story. You also need to know whether your brand is appearing inside the AI's answer in the first place.
Why Reddit and YouTube matter more than you'd think
One of the more surprising data points: in a study of Google AI overview citations, YouTube made up 14% of sources and Reddit contributed around 7%, together accounting for roughly a fifth of all citations.
Why? AI models aren't just pulling from polished company websites. They're also looking for authentic, first-person content: people explaining their situation, comparing options, or sharing what actually happened to them. In an analysis of 250,000 Reddit posts cited by GPT Search, Perplexity and Google AI mode, Q&A threads, comparisons, personal stories and discussion posts made up nearly 75% of what got cited.
The takeaway isn't "go spam Reddit." It's that genuine participation in communities where your buyers are already discussing their problems can genuinely feed into how AI recommends your brand.
Five technical basics to check today
Lucia ran through some foundational, easy-to-action items every founder should look at:
- llms.txt – a simple text file summarising what your business does, linking to your most important content. Not universally required (Google's said its AI search doesn't need it) but low-effort to add.
- robots.txt – controls which crawlers can access your site. Worth checking you haven't accidentally blocked a relevant AI crawler.
- Sitemap – an up-to-date map of your pages, so AI systems can find new and updated content easily.
- Noindex tags – make sure these aren't accidentally applied to pages you actually want found.
- JSON-LD schema – machine-readable labels that help AI understand what a page represents (a product, an article, an author, and so on).
None of these are one-off fixes. As Lucia put it, GEO isn't something you configure once and forget. Platforms are constantly changing how they crawl, cite and rank content, so a brand can become more or less visible without changing anything itself.
Measuring what's actually happening
Here's the tricky part: Google Search Console and Analytics only show human clicks and visits. They don't show you every time an AI bot reads your site.
Lucia broke the measurement problem into three separate questions:
- Is AI even crawling your site?
- Is it being cited in AI answers?
- Are humans actually clicking through from those citations?
Being crawled doesn't guarantee being cited. Being cited doesn't guarantee a click. Each needs to be tracked separately, and ideally by platform, since each AI selects and weighs sources differently.
SEO is still the foundation, but it's no longer the whole picture. The goal isn't to chase every algorithm update (an impossible task anyway) but to make your brand easy to access, easy to understand, easy to trust, and possible to measure across the platforms your buyers are actually using.