Alvaro Peña de Luna 
I asked the four big AI engines (ChatGPT, Gemini, Claude and Perplexity) the same question 320 times: which is the best marketing agency in Spain?
Not once. 320 times, with web search enabled, half in Spanish and half in English, across 20 real buyer questions. Then I measured who the AI actually recommends.
The ranking matters less than what the data reveals about how AI search decides who to recommend. If you sell anything online, this should worry you.
What AI recommends today
Share of Voice here is the percentage of the 320 responses where each agency appears. NeoAttack leads at 39%, followed by Elogia (36%), Dobuss (30%), iSocialWeb (28%), Flat 101 (27%) and Making Science (26%).
That is the headline. The interesting part is everything underneath it.
Finding 1: every AI recommends a different agency
There is no “best agency according to AI”. There are four rankings, and they barely touch.
- ChatGPT crowns Brandevs and La Teva Web, with iSocialWeb third.
- Claude puts NeoAttack, Elogia and Flat 101 on top.
- Gemini surfaces names that almost no other engine mentions, like ExpandX and Algofy.
- Perplexity hands the win to Dobuss by a wide margin.
Winning in one engine tells you nothing about the others. A brand that is everywhere in Claude can be a footnote in Perplexity. If you only check ChatGPT, you are measuring a quarter of the picture.
Finding 2: language changes everything
The same question in Spanish and in English produces two different worlds.
Dobuss dominates in Spanish (53%) and nearly vanishes in English (8%). In English, the AI forgets Spanish agencies and starts recommending US firms like WebFX, NP Digital and Single Grain. Making Science, an international and listed company, actually gets stronger in English.
Your AI visibility depends on the language of whoever is asking. A Spanish prospect and an English-speaking prospect see different shortlists.
Finding 3: AI does not cite your website
This is the finding that reframes the whole game. When the AI recommends an agency, where does it get the information?
Not from the agencies’ own websites. The most cited sources are:
- Directories: Sortlist, Clutch, Semrush Agencies.
- Media and rankings: El País, marketing4ecommerce, puromarketing, Escuela Europea de Empresa.
Only a handful of agencies get their own site cited as a source, and they are the ones that have clearly invested in content (Dalma Media, AREA10, Idital). For everyone else, the AI mentions them because a third party ranked them, not because of their homepage.
The takeaway: being mentioned is not the same as being cited. Your best landing page does not make you visible in AI. Being present where the AI looks does.
What this means
Classic SEO does not guarantee that AI recommends you. Showing up in ChatGPT or Perplexity is a different discipline, with different signals: mentions, directories, third-party rankings and the language of the query. That discipline is called GEO, Generative Engine Optimization.
And it is measurable. Per engine, per language, per source. That is exactly what we built Mencoro to do: it tells you whether AI recommends you or your competitors, in each engine, and where that recommendation comes from.
Want to know if AI recommends your brand? Try the Mencoro demo, no signup required.
