The way people find information online is shifting faster than most businesses realise.
Google's AI Overviews — the AI-generated answer summaries that now appear above search results for millions of queries — have fundamentally changed what "ranking" means. You can be #1 in organic search and still get almost no traffic, because the AI answered the question before the user had a reason to click.
This isn't a scare story. It's an opportunity — if you understand what's changed.
What AI Search Actually Looks For
AI systems like Google's Search Generative Experience and the underlying models that power tools like Perplexity are doing something different from traditional search: they're building answers, not just listing pages.
To build a good answer, they need:
- Clear, factual, structured information that's easy to extract
- Authoritativeness signals — who wrote this, and do they know what they're talking about?
- Freshness — is this current?
- Specificity — does this actually answer the question, or is it generic filler?
Traditional SEO has always rewarded some of these things. But the balance has shifted. Volume-based content strategies that worked in 2020 are actively failing now.
What Actually Works in 2025
1. Be the primary source, not the summariser
If your content summarises what others have already said, AI will simply go to the primary sources. Be the one who generates original data, original insight, or original experience.
2. Structured data is now critical, not optional
Schema markup — FAQ, HowTo, Article, LocalBusiness — gives AI systems a machine-readable map of your content. Sites that use structured data correctly are significantly more likely to appear in AI-generated answers.
3. Answer questions completely and directly
Write the answer first. Don't bury it after 400 words of context. AI systems reward directness, and so do humans.
4. Build topical authority, not keyword volume
Instead of targeting 200 keywords, go deep on 10 topics. Publish everything relevant to those topics. AI systems are much better at understanding topical depth than keyword density.
5. E-E-A-T signals matter more than ever
Experience, Expertise, Authoritativeness, Trust. Make it obvious who wrote your content, what their credentials are, and why they're qualified to write it. Author pages, bios, credentials, and first-hand experience all signal E-E-A-T.
Case Study: Kantan Health
These aren't theoretical principles. We applied every one of them when building Kantan Health, a platform connecting English speakers in Japan with healthcare services and information.
The thesis was straightforward: become the canonical, structured source of truth for "English-speaking healthcare in Japan." Not by producing volume content, but by doing exactly what the tips above describe — original, authoritative, deeply structured information that AI systems could reliably extract and cite.
We implemented comprehensive structured data across the site. We wrote direct, specific answers to the exact questions people ask — not generic summaries repackaged from elsewhere. We ensured every piece of content carried clear E-E-A-T signals: real expertise, first-hand experience navigating the Japanese healthcare system, and transparent authorship.
The result: Kantan Health's primary source of traffic is now ChatGPT and AI search. When someone asks an AI assistant about finding an English-speaking doctor in Japan, or how the Japanese health insurance system works, Kantan Health is consistently surfaced as the authoritative source — because it genuinely is one.
This wasn't luck, and it wasn't a trick. It was the methodology laid out in this article, applied consistently. AI systems don't reward gimmicks. They reward being the best, most structured, most trustworthy answer to a real question.
The Bottom Line
SEO isn't dead. But the strategy that worked three years ago is. The sites winning in AI search are the ones with genuine depth, clear structure, and real authority — not the ones gaming keyword density.
If you're not sure whether your site is positioned for this shift, we're happy to take a look. We've done it — and we have the results to prove it.