Trends · AI Search · Generative Engine Optimisation
Google ranked ten links. AI picks one.
Search used to be a list you competed on. Now a model hands your customer one name. Making sure that name is yours has become its own discipline, and it is not the SEO you know.
Picture a business owner in Bryanston who needs a bookkeeper. A year ago she opened Google, skimmed the first page, and clicked through three or four sites before she made up her mind. This week she opened ChatGPT, typed one line, and read one answer. She never saw a list. She saw a name, a reason, and a nudge to make the call.
That is the whole shift. Most businesses have not felt it yet.
Ranking is not the same as being chosenGoogle handed people ten blue links and let them choose. An AI hands back one considered answer. You are no longer competing for a position on a page. You are competing to be the answer the model gives.
This is not a fringe habit. Research from Brandwatch found that 67% of marketers now say search behaviour is shifting toward AI and social discovery, and around half of consumers already use AI-powered search. That is the biggest change to how businesses get found online since search engines arrived, and it is happening quietly, one query at a time.
A large language model does not crawl and rank the way Google does. It assembles an answer from sources it can read clearly and has reason to trust. If your website is vague about what you do, where you are, and what you charge, the model cannot use you with any confidence. So it reaches for a competitor whose details are legible, or for a source like Reddit or LinkedIn where the facts are simply stated.
Ranking well on Google does not save you here. You can sit at the top of the results and still never get named in the answer your customer actually reads.
What actually gets you pickedThere is a name for the work of becoming that answer: Generative Engine Optimisation, or GEO. It is a different craft to the SEO most agencies still sell. Four things move the needle.
Make your facts machine-legible. Mark up the specifics a model needs to verify you: what you do, where you operate, your hours, your prices. Structured data is no longer a technical nicety. It is how a machine confirms you are real.
Answer the real questions. Build pages around the questions your customers actually ask, in their words, not around keywords. If someone asks an AI who does this work in Sandton, your site should already carry that answer plainly.
Prove who is behind the work. Clear authorship, real sources, and consistency across your site, your images and your captions. Models weigh trust, and anonymous, generic content reads as risk.
Show up where models read. Reddit and LinkedIn are among the sources AI models cite most. A genuine presence there is now part of being findable, not a side project.
These signals now matter as much as keywords ever did. Get them right and you stop hoping to be found. You become the thing the machine reaches for.
The window is open, and it is narrowHere is the part worth acting on. Almost nobody in the local market is doing this yet. The businesses that move first become the default answer in their category, and a default is expensive to unseat. First-mover advantage in a new channel is real, and it does not stay open for long.
In 2026, visibility is about being selected, not surfaced.
At Ten 15 J.T., we treat this as its own discipline. We ask the AIs the questions your customers ask, show you where you appear and where a rival appears instead, and put the signals in place that make you the answer. Not more content. The right answer, in the one place your customer now looks first.