AI SEO vs traditional SEO: what actually changes

A growing share of searches now end without a click. Someone asks ChatGPT or Perplexity, or Google shows an AI Overview, and the answer arrives without anyone visiting a website.

That creates a second job alongside ranking: being the source the model quotes.

The distinction, precisely

Traditional SEO aims to place a page in a list of results. Success is a position.

AI SEO — sometimes called Generative Engine Optimization — aims to have your content understood, trusted and cited inside a generated answer. Success is a mention, often with a link.

They overlap heavily. Much of the technical groundwork serves both. But the optimisation targets differ, and a few things matter for one and not the other.

What's the same

Don't let anyone sell you a completely separate discipline. These still matter and always did:

• Crawlability. If it can't be read, it can't be cited.
• Page speed. Slow pages get crawled less.
• Clear heading structure. Helps both parsers and models.
• Genuine authority. Models weight sources that other sources reference.
• Accurate, specific content. Vagueness was never rewarded and is now actively useless.

If your technical SEO is broken, fix that before thinking about AI. There's no AI-specific shortcut past a site that can't be crawled.

What's genuinely different

1. Answer-first structure.

Models extract statements. A self-contained, declarative sentence is liftable. A conclusion that depends on three preceding paragraphs is not.

Put the answer first, then support it. This is the opposite of the build-up-to-a-reveal structure most content marketing uses, and it's the single biggest writing change.

2. Entity clarity.

A model needs to know what you are before it will cite you. Structured data — Person, Organization, Service — states it explicitly rather than making it inferable.

Consistency across the web matters too. If your business details differ across your site, your Google Business Profile and your LinkedIn, the entity resolves less confidently.

3. Question-and-answer formatting.

FAQ sections written as real questions, in the phrasing people actually use, with self-contained answers. This maps directly onto how models retrieve and quote.

Not "Why choose us?" — nobody asks that. "How much does X cost in South Africa?" — people ask that constantly.

4. llms.txt.

A plain text file at your root summarising what you do, in a format that's trivial to parse. It's a convention rather than a standard, and I'd be overstating things to claim proven impact. It costs twenty minutes.

5. Being referenced elsewhere.

Models draw on the whole web. Being mentioned on other reputable sites influences whether you're seen as a credible source — arguably more directly than it influences rankings.

What I can't yet tell you

Whether any of this materially moves citation rates, with data.

Measuring AI citation is genuinely difficult. There's no Search Console equivalent telling you how often ChatGPT mentions you. Tools claiming to measure it are estimating.

I've implemented all of the above on this site. I don't have clean before-and-after data, and anyone showing you a confident case study on this right now is probably selling something.

Why I'd do it anyway

Because the work is defensible under both outcomes. Structured data, clean headings, answer-first writing and real FAQs improve traditional search regardless of what happens with AI.

That's the test I'd apply to any AI SEO recommendation: would this still be worth doing if AI search stalled tomorrow? If yes, do it. If no, wait for evidence.