How ChatGPT chooses which estate agents to recommend

Ask ChatGPT “who is the best estate agent in Marbella for British buyers?” and you get a short answer with two to four names. Ask Gemini or Perplexity and you get a similar list, sometimes with different names. For the agencies on that list, it is free, high-intent demand. For everyone else, it is invisible lost business.

So how does ChatGPT decide?

It doesn’t have a ranking. It has an impression.

There is no hidden league table of agents. A language model builds its answer from what it has read about each agency, plus, in many cases, a live web search at the moment you ask. It then names the agencies it can describe most confidently for that specific question.

That word confidently is the key. AI engines prefer businesses they can say something clear and true about: “English-speaking, strong on the Golden Mile”, “specialises in new builds in Benahavís”. An agency with a vague website and generic reviews gives the model nothing to repeat.

Where that impression comes from

The sources AI engines draw on for Costa del Sol property questions fall into five groups.

1. Your own website

AI engines read your site, especially pages that answer buyer questions directly. An English-language guide to buying in Mijas, a page on how you help Dutch buyers, or a clear “about us” that says who you serve all give the model something to quote.

What hurts: vague copy (“we are passionate about property”), key facts hidden in PDFs or images, and pages that only exist in Spanish.

2. Reviews

Reviews on Google, Trustpilot and the portals carry a lot of weight. The detail matters more than the star rating. A review that says “they helped us buy in Nueva Andalucía from London, all in English” tells the model exactly which questions to name you for.

3. Press and expat media

Local English-language media such as Sur in English, Euro Weekly News and The Olive Press are frequently cited in AI answers about the coast. So are relocation guides and “moving to Spain” articles. One good feature can do more than a year of social posts.

4. Portals and directories

Your listings on Idealista, Kyero, Rightmove Overseas and local directories confirm you exist, where you operate and what you sell. Inconsistent names or addresses across them make the model less sure who you are.

5. Structured data

Schema markup on your site, such as RealEstateAgent with the areas you serve and the languages you speak, gives AI engines machine-readable facts. It’s rarely the deciding factor alone, but it removes guesswork.

Why the big names don’t always win

You might expect the biggest agencies to dominate. Often they don’t. A smaller agency with a sharp specialism (“the agent for Scandinavian buyers in Mijas”), recent detailed reviews and one or two press mentions regularly gets named ahead of larger firms with more stock.

That is good news. AI visibility rewards clarity and evidence, not budget.

Five things to do this month

1. Ask ChatGPT, Gemini and Perplexity the questions your buyers ask. Write down who gets named. That is your baseline.
2. Say who you are for, in plain English, on your home page. Area, buyer type, languages.
3. Ask your last ten happy buyers for a review that mentions where they came from and where they bought.
4. Check your name, address and description match across your site, Google Business Profile and the portals.
5. Write one genuinely useful English guide for the buyers you want most, such as “buying in Estepona from Norway”.

Want the full picture?

Our free AI visibility check runs 10 real buyer questions through ChatGPT, Gemini and Perplexity and compares your agency with three competitors. You get a short report within five working days.