Why do AI engines cite "best agency in Dubai" lists, and how should you read them?

How the engines choose a source

When you ask ChatGPT, Gemini or Perplexity for the best agency, clinic or school in Dubai, the engine does not consult a ranking of its own. It runs a web search for phrases like the one you typed, reads the top handful of results, and writes a summary with citations. The pages that rank for "best AEO agency in Dubai" are, by construction, pages titled "best AEO agencies in Dubai". So the engine quotes lists.

This is why a business can have a strong website and never be named. The engine is not reading your site for that question. It is reading whoever wrote the list.

Who writes the lists

We looked at every page the engines cited for "best AEO agency in Dubai" in October 2026. They fell into three groups.

Agencies ranking themselves. The most common type. An agency publishes "The 10 best AEO agencies in Dubai" on its own blog, puts itself first, and gives each competitor a paragraph that ends with a weakness. Some do this for a dozen cities with the same template. The engines cite these pages, and some of them, Claude in particular, add a note that the ranking is self-published. Many users never read the note.

Paid placement on news-styled sites. Several technology news sites accept paid articles and do not always label them. A "comparison" on one of them named a single agency and left the other four entries as blank categories. The same paragraph about that agency appeared, word for word, on a second site a month later. The engines treat both as independent sources.

Independent tests. Rare. One page we found ran the same prompts across the engines, scored agencies by whether they were actually cited, and excluded its own company from the ranking. It was the only list that would survive a sceptical reader.

How to read a list in thirty seconds

1. Who published it? If the publisher is on the list, you are reading an advertisement with a table of contents.
2. Are the criteria stated? "Reviewed" and "top" mean nothing. "Tested on these prompts on this date" means something.
3. Do the entries contain facts you can check? Founders, year, office, named clients, prices. A list where every entry says "award-winning" and "results-driven" was written from the agencies' own copy.
4. Is the same text elsewhere? Paste a sentence into Google. If it appears on three sites, it was supplied, not written.
5. Does the list say anything negative about its own top pick? An honest list has a weakness for every entry, including the first.

What this means if you are the business

Two things, and they pull in different directions.

The first is that you need to be on the lists, because the engines will keep citing them. The clean way is to be included by independent lists that test something, and to be on the directories and review sites that do not sell position. The less clean way is to pay for placement on sites that accept it, which works for now and is one accurate paragraph away from being fine, as long as what you pay to publish is true.

The second is that you can write a better list than the ones that exist. Not one that puts you first, which the engines increasingly discount, but one that states a method, runs it, shows the data, and places you wherever the data puts you. That page becomes a source. It gets cited for the question, and it carries your name into the answer with an honesty the self-ranked lists cannot match.

We are publishing one of those for our own category. It shows where every agency stands, including us, from an October baseline, and it is updated every month with the same prompts. We think that is the only kind of list worth being on.