What International Buyers on the Costa del Sol Actually Ask AI Before Buying

Why people ask an engine what they never typed into a portal

Because the interface changed what feels askable. A portal search is a filter — bedrooms, price, area — and it presumes you already know you can and should buy. A conversation with an engine is private, unlimited and unembarrassing, so it collects the questions people are not ready to put to an agent whose income depends on the answer.

That produces a genuinely different question set, arriving months earlier in the decision. By the time a buyer contacts anyone, the research that formed their view has already happened somewhere the seller never saw.

The five categories of question

Grouped by what the buyer is actually trying to resolve rather than by topic. Nearly everything asked before a first viewing falls into one of these.

Notice what is absent: almost none of it is about inventory. The decision is being made about a jurisdiction and a process long before it is made about a property.

What the portals answer, and what they don’t

Portals answer the inventory question extremely well and the other five categories barely at all — and the reason is structural rather than lazy. A portal’s asset is listings and its incentive is to move a visitor toward one. Tax, legality and residency answers are jurisdictional, change over time, carry real liability if wrong, and move the buyer away from a listing while they think.

Agencies have a version of the same problem: the answers are commercially inconvenient, and writing them down publicly invites scrutiny. So the highest-influence questions in the whole purchase are left to forums, expat comment threads and outdated blog posts — the thinnest possible source layer for the highest possible stakes.

Why English-language sources are wrong here

Because every question above is jurisdiction-specific, and a generic answer is not merely thin — it is frequently wrong in a way that costs the buyer money. Transfer tax differs by autonomous community. Rental licensing differs by municipality. What counts as a legally buildable rural property in Andalusia is not what it is elsewhere.

This is the clearest real-world instance of the boundary set out in the citation arbitrage piece: for language-independent knowledge, an engine can translate a foreign source and be right; for a jurisdictional question it cannot, because there is no correct answer that is not local. And the awkward part is that the buyer is usually asking in a third language — German, Dutch, Swedish — about a Spanish jurisdiction, so neither the sources in their language nor most sources in Spanish are positioned to answer.

What this means for anyone selling here

1. Stop competing on inventory in search. Portals will always have more listings, better feeds and more domain authority for property-shaped queries. That competition is already lost and was never worth entering.
2. Compete on process authority. The five categories above are contested by almost nobody with genuine local standing, and they are what the buyer is researching in the months when they are still deciding whether to buy at all.
3. Publish the inconvenient answers. Total cost including every fee. What can go wrong with an off-plan purchase. When a rural property is not legally buildable. The credibility of saying so plainly is the moat, and the reluctance of competitors to say it is why the moat exists.
4. Date and localise everything. Name the autonomous community, name the year, state when the rule last changed. An undated tax answer is unusable to an engine trying to be correct.

How to build the answer layer

• One page per question, answered in the first two sentences, then the detail and the caveats — the construction described in answer-shaped content.
• Per-language editions, not translations. A German buyer’s question about Andalusian purchase tax needs a German-language answer written for a German buyer’s situation, not a rendered Spanish page.
• A named, credentialed author. On legal and tax adjacent content, an anonymous page is discounted by readers and engines alike.
• Real numbers with dates attached. Percentages, thresholds, typical ranges, and when each was last verified.
• Measure with a prompt panel. Run the real questions monthly and log who gets cited, exactly as set out in the GEO audit.

The honest limit of this piece

This is a taxonomy of the question space, not a measured study. It is reasoned from what the purchase actually requires a foreign buyer to resolve, and from running these questions against engines ourselves — not from a dataset of query volumes, which for private conversations with assistants does not publicly exist.

We are stating that plainly because the alternative is the thing this category is full of: confident percentages with no source. If you want the version of this specific to your market and language, the method is the prompt panel above, and it takes an afternoon to start.

How we think about it at Thinktier

SolProp exists in this gap — helping international buyers understand, evaluate and access Costa del Sol opportunities, which in practice means answering the five categories rather than listing more properties. It is a distribution-first business by construction: the answer layer is the asset, and the transactions follow the trust it earns.

The reason we think this position is defensible is not that the content is hard to write. It is that the incumbents are structurally disincentivised from writing it — and a competitor who is unwilling to answer the buyer’s real question cannot take the position from you by spending more.