For twenty years, being found meant ranking. Someone typed a query, saw ten links, clicked one, and you could measure the whole thing. That still happens and it still matters.
What is new is that a growing share of buyers never see that page. They ask an assistant, get an answer assembled from sources the model decided to trust, and act on it. No ten blue links, often no click at all. If you are not among the sources, you are not in the conversation, however well you rank.
These are two different problems. They overlap, but they are not won the same way, and a supplier who talks about one while quietly ignoring the other is selling you half the work. Both are what my search and AI visibility work covers.
What has not changed
Classic search still rewards the same things: a site that can be crawled quickly, pages built around what a person is actually trying to find out, internal linking that makes the important pages obviously important, and — locally — an accurate business profile with consistent details.
One local caveat worth stating plainly: putting “Dubai” in a heading is not local relevance. Genuine local signals are addresses, real regional references, work that demonstrably happened here, and consistency across every place your business is listed.
What is different about being quoted
Getting lifted into an answer box rewards a specific shape of content: the answer near the top rather than buried after eight paragraphs of preamble, structured data that states plainly what the page is, and claims a machine can corroborate somewhere other than your own site.
The preamble habit is the single most common problem I see. A page that opens with three paragraphs of throat-clearing before answering the question will lose to a page that answers it in the first sentence, even if the first page is better written.
What is different about being recommended
Being named by an assistant is closer to reputation than to ranking. Models draw on what they can find about you across the whole web, not just on your own site. So the work splits in two.
On your site: clear entity information, consistent naming, and structured data that connects you to your subject rather than just describing a page. Off it: being cited, listed and mentioned in the places the model already trusts — trade bodies, real directories, press, other people’s writing.
Nobody can guarantee an assistant will name you, and anyone promising that is guessing. What you can do is make yourself the easiest candidate to describe accurately.
Entities, not keywords
Both search engines and language models increasingly reason about things rather than strings: this company, this person, this service, in this place. Getting that right means saying clearly and consistently who you are, what you do and where — in a form machines read.
It is unglamorous work: structured data (the vocabulary is at schema.org), matching profiles, the same facts stated identically everywhere they appear. It is also the highest-leverage work on most sites I audit, because so few businesses have bothered.
The bilingual part, which almost everyone gets wrong
Arabic and English are separate search markets with different competition, different phrasing and different intent. A page translated word for word usually targets nothing in either language, because nobody searches in translationese.
Arabic search deserves its own keyword research and often its own pages, not a mirror of the English ones. In a market where a large share of buyers search in Arabic, treating it as a translation layer is leaving the easier half of the opportunity on the table — and it shows in the design as much as in the rankings, which is the subject of Arabic-second always shows.
How to tell whether it is working
Rankings are a proxy and an increasingly weak one. Measure enquiries, calls, and qualified traffic. Separately, ask the assistants themselves — periodically put your own buying questions to them and see who gets named. If the answer is nobody in your category, that is an opening rather than a problem — the same one that makes unfinished AI projects so common. It is crude, it is not a dashboard, and it tells you more than a rankings screenshot that improves while the phone stays quiet.