What does “AI SEO” mean?
Two different things, and the honest one is the less exciting one. The dishonest version is a tool that produces five hundred pages of keyword-shaped text overnight. Google’s guidance on AI-generated content is explicit that using automation to generate content primarily to manipulate rankings is a violation of its spam policies, and that its systems reward quality “regardless of how content is produced” 1. The honest version is using the tools to do the SEO work faster: research the questions people actually ask, draft, structure, check, and then have a person sign it.
What does Google reward now?
The list has not changed much; the weighting has. A page that answers the query, near the top, in plain words. Headings that are the questions. A named author with a reason to know. A publication date and a revision date that actually moves. Evidence: a number with a source, a source with a link. And a page that loads: Google’s own research found 53% of mobile visits abandoned when a page takes longer than three seconds 2, and a 0.1-second improvement in mobile speed lifted lead-generation page bounce rates by 8.3% in Deloitte and Google’s 30-million-session study 3.
What is different about AI search?
It quotes. An answer engine lifts a passage and attributes it, so the page that gets lifted is the page structured to be lifted: a direct answer, a question-form heading with the answer underneath, a table (a row can be quoted; an argument cannot), a statistic with a linked source. Being cited by an assistant is the new version of ranking, and it rewards clarity and evidence more than the old version did.
| Element | Why it ranks | Why it gets cited |
|---|---|---|
| Short answer at the top | Matches the query | Liftable in one passage |
| Question headings | Matches how people search | Maps to the question asked |
| A data table | Structured content | A row is quotable |
| Linked sources | Trust signal | Checkable, so trusted |
| Author, date, revision | Freshness and expertise | Attribution |
| Fast on mobile | Core Web Vitals | Crawled fully |
Where does AI SEO go wrong?
Invented evidence. A model asked to make a page persuasive will produce statistics, studies and quotations that do not exist, and a page that publishes them has made a checkable false claim under the business’s name. The rule this blog now follows is blunt: no page may say it surveyed 1,000 businesses unless somebody surveyed 1,000 businesses, and every figure carries a link somebody opened. Google’s own advice points the same way: focus on accuracy, and consider telling readers how AI was used 1.
Does it work, then?
Yes, when “it” means faster research, drafting and structuring, and a person checking, citing and signing. No, when “it” means volume. The pages that rank and get cited in 2026 are the ones a careful person would have written with more time; the tools give the time back.
What should a South African business do?
Start with the site. A fast, hand-built site with the answer on the page is the base everything else sits on; that is what the website design product delivers, with the copy, the structure and the search markup included. Then write the pages that answer the questions your customers ask, in their words, with the numbers checked.
Questions people ask
Does Google penalise AI-written pages?
Not for being AI-written. It acts against content made primarily to manipulate rankings, however it was produced, and rewards useful content, however it was produced 1.
How do I get cited by AI search?
Answer the question at the top of the page, use question headings, add a table and link your sources. Assistants lift what is easy to lift and trust what is easy to check.
Is page speed still an SEO factor?
Yes, and a conversion factor before that: 53% of mobile visits abandon a page slower than three seconds 2.
Sources
- Google Search Central, “Google Search’s guidance about AI-generated content”, February 2023
- Google / SOASTA, “The Need for Mobile Speed” (2017): 53% of mobile visits are abandoned if a page takes longer than three seconds
- Deloitte Digital with Google, “Milliseconds Make Millions” (2020) — 30 million sessions across 37 brands