SEO isn't dead — it's evolving. That's a thought I keep coming back to as I watch the market share of traditional search engines like Google slipping, while new AI entrants like Perplexity and DeepSeek quietly (and sometimes not so quietly) eat their lunch.
I was skeptical about AI search at first. Getting people to change a deeply ingrained process — like turning to Google to look up information — takes time. And Google can always co-opt other AI solutions by baking AI into their own platform, as they've done with AI Overviews.
But AI search engines are growing. People like them. And so, for anyone running a business, it's becoming more and more important to be the brand those AIs recommend.
The shift in numbers
Per a recent SparkToro report, U.S. desktop search market share looked like this in mid-2025:
- Google — still dominant, but down ~6 points from 2023
- ChatGPT search — fastest-growing standalone search interface
- Perplexity — small but committed user base, growing month over month
- Gemini, Claude, Copilot — built into other surfaces; harder to isolate but clearly being used
The directional story is more important than the absolute numbers. Search is fragmenting. Where it used to be 1 search box, it's now 5 — and the AI-native ones don't show a list of links. They show one answer.
What “ranking” means in AI search
Google's old game: you write a page, you optimize it for a keyword, the page either ranks or it doesn't. You can A/B test variations, see your rank, iterate.
AI search's new game: someone asks a question. The AI synthesizes a 200-word answer mentioning 2 to 5 specific brands. You either are one of those brands or you're not. Position 6 might as well be position 600 — the user already has their answer.
That makes citation (yes / no, name / no-name) the primary metric, and position-when-citedthe secondary one. The classic “rank for a keyword” SEO mindset doesn't map cleanly.
What still works
Most traditional SEO fundamentals still matter, because AI engines train on the web and search the web:
- Authority and citations. Wikipedia mentions, news coverage, high-DR backlinks — all still weight heavily in what AIs surface.
- Schema markup. JSON-LD is now arguably more important than it ever was for Google, because AIs use structured data to disambiguate brands.
- Clear, declarative content. Pages that answer specific questions in specific words rank well in AI answers. Filler hurts.
What's new
What's actually different is the kind of content that gets cited:
- Comparison pages (“X vs Y”) get quoted in product recommendation queries far more than category pages.
- FAQ blocks get verbatim-quoted by Gemini and Claude more than any other content type.
- Independent review citations (TechCrunch, Reddit, niche newsletters) matter more than they did for Google.
What to do this week
The single highest-leverage move: run a free AI visibility scan on your own domain so you know where you actually stand. Without that, every optimization decision is guesswork.
From there, the playbook is straightforward: pick the 3 highest-impact gaps, fix them, re-scan in 30 days. AI engines update their citation patterns surprisingly fast — improvements show up in 2 to 4 weeks for most categories.
AI search isn't coming. It's here. The brands that show up in the AI answer become the brands that win. Period.