ChatGPT cites a higher proportion of .es domains than AI Mode or Gemini in the responses for Spain examined by Aleyda Solis . But the ranking changes when references to Google and YouTube are removed: AI Mode then moves into first place. The difference reveals how much a visibility reading depends on the sources included in the comparison.
The study , published on 8 October 2026, places the .es citation share at 24.8% for ChatGPT, 22.6% for AI Mode and 16.6% for Gemini. The sector contrast is larger still. .es domains account for a majority of citations within the personal finance and home topics; health, beauty and business software show much smaller proportions.
Solis, a search optimisation consultant associated with Orainti , compared the sources used by the three platforms across ten topics and four markets: the United States, United Kingdom, Spain and Brazil. She uses the same library of questions localised for each market. Her research shows what changes when a response seeks information for a different audience, even when the AI product is the same.
For Spanish companies working on acquisition through content, the question has an immediate consequence: competition to appear as a source is organised by country, topic, platform and page. A brand can share that space with public bodies, publishers, retailers and service businesses. The study separates these layers and shows why a national average explains only part of the available visibility.
What a local source means in this study
The research uses country domains to identify sources targeting each market: .es for Spain, .uk for the United Kingdom and .br for Brazil. Solis selects them as a consistent geographic signal. The category called local describes this classification rule, allowing the same approach to be applied across the three countries.
A Spanish company may publish on a .com domain, just as a publisher or service may use a generic domain to address Spain specifically. Those pages enter the study's non-local group. Consequently, the .es percentage does not represent the proportion of Spanish companies, nor does it automatically divide responses between Spanish and foreign content.
That distinction is particularly relevant when interpreting the 24.0% share the report publishes for .es citations in the overall Spanish results. The remainder may include Spanish-language pages and localised material. The finding describes a geographic extension's presence among cited sources and creates a comparison of how that presence is distributed.
The unit is also specific: citations used in the responses examined. A citation shows that the platform included a reference in its answer. Visits subsequently received by that page and the reader's actions belong to another moment in the acquisition journey. The study concentrates on source selection, allowing that space to be examined on its own terms.
Share of .es domains in citations for Spain
Proportion of references to .es domains in responses for Spain from ChatGPT, AI Mode and Gemini, in Aleyda Solis’s source comparison published on 8 October 2026.
Tap or hover for details. All values are also available in the data table.
Collapse notes
The .es extension is the study’s geographic indicator; generic domains can also publish content targeting Spain. Citations, rather than users or clicks.
Explore the data table
| Category | Citations from .es domains |
|---|---|
| ChatGPT | 24.8% |
| AI Mode | 22.6% |
| Gemini | 16.6% |
Finance and home change the Spanish average
In Spain, personal finance reaches a 59.4% .es citation share and home and DIY 52.9%. These are the two topics in which the extension represents a majority. Automotive stands at 38.0%, while travel is at 23.2% and fashion and clothing at 18.4%.
At the other end, health and wellbeing records 2.8%, beauty and skincare 9.4%, and B2B software and SaaS 10.5%. Consumer electronics accounts for 12.2%, and sport and fitness for 12.8%. The distance between categories turns the average into a starting point requiring a sector-level reading.
For a content team, these results describe different settings. A source competing in a finance question may have a specific national context. In software questions, the reference set more often includes domains outside the local ccTLD. Identifying the pages appearing in each topic helps reveal which information the response uses to address it.
The same general pattern appears in the other countries, with differences. Personal finance, home and automotive occupy the top three positions by local share in all three markets. However, home leads the British ranking at 66.3%, while finance leads in Spain and Brazil. The hierarchy changes even among topics with the strongest country-domain presence.
B2B software provides a particularly consistent comparison. Its local share is 14.3% in the United Kingdom, 10.5% in Spain and 15.9% in Brazil. The result places category visibility within a source selection predominantly outside the local ccTLD across all three markets, without turning that pattern into a judgment about the quality of the products or companies cited.
The .es presence changes across ten topics
Share of citations to .es domains by topic in the responses for Spain studied by Aleyda Solis. Categories compare the country extension’s weight within each topic set.
Tap or hover for details. All values are also available in the data table.
Collapse notes
Percentage of citations within each topic, rather than market allocation across sectors. A .com localised for Spain remains in the study’s non-local group.
Explore the data table
| Category | Citations from .es domains |
|---|---|
| Personal finance | 59.4% |
| Home and DIY | 52.9% |
| Automotive | 38.0% |
| Travel | 23.2% |
| Fashion and clothing | 18.4% |
| Sport and fitness | 12.8% |
| Consumer electronics | 12.2% |
| B2B software and SaaS | 10.5% |
| Beauty and skincare | 9.4% |
| Health and wellbeing | 2.8% |
Travel and fashion show another competitive scale
The country table shows which part of Spain's difference remains when the same topic is compared. In travel, the country-domain share is 23.2% in Spain, 35.6% in the United Kingdom and 43.4% in Brazil. All three markets have a majority of references to domains outside the local ccTLD, but the proportions differ. Classification by extension reveals a different source composition in front of each audience, even within the same industry.
In fashion and clothing, Spain records 18.4%, against 34.4% in the United Kingdom and 48.8% in Brazil. In the latter market, country-domain references approach half the total. The contrast shows why a brand developing content across locations needs to examine the localised response. The same commercial category can encounter a different source mix in each destination, making a country-level reading relevant to the content being prepared.
Consumer electronics provides another comparison: Spain's 12.2% sits alongside 20.2% in the United Kingdom and 30.2% in Brazil. Within Spain, Gemini also has the highest .es proportion of the three platforms in this topic. The topic measure and the platform measure provide complementary readings. One describes the sector's general source mix; the other locates where that mix changes within the tools examined by the research.
For acquisition analysis, retaining both dimensions helps formulate a specific reporting question: what information do the selected pages provide for that market? The same generic domain can host different material by country. The US comparison, where page overlap is lower than site overlap, reinforces the value of examining the content that appears rather than stopping at the publisher's name. It gives the content team a more precise object to study.
The research organises its questions through a library localised for the markets being compared. That choice connects the tables with search situations equivalent in their underlying approach, instead of setting entirely unrelated topic lists against one another. The result makes it possible to examine how source selection changes when a question is taken into another context, while retaining the topic and distinguishing the platform supplying the answer. This is the research structure behind the commercial interpretation, rather than a separate claim about how shoppers behave.
A platform can lead in one topic and trail in another
ChatGPT leads the .es percentage in the overall Spanish results, but the topic breakdown changes the picture. In personal finance it reaches 62.7%, against 30.4% for AI Mode and 22.4% for Gemini. In home and DIY, the proportions are 54.8%, 30.6% and 29.9%, respectively.
The difference moves in beauty. AI Mode records a 23.7% .es share, while ChatGPT is at 8.3% and Gemini at 10.8%. In travel, AI Mode also has the highest figure, at 29.4%, compared with ChatGPT's 23.4% and Gemini's 18.7%. Aggregate leadership does not repeat in the same way across every topic.
Consumer electronics supplies a third example: Gemini reaches 17.4%, above AI Mode's 12.0% and ChatGPT's 11.4%. The comparison avoids attributing a uniform local preference to any platform. What appears in the responses depends on the combination of market and subject being discussed.
This variation changes how a content opportunity can be described. The question is more specific than which AI product cites more .es sources. It also concerns which pages are used when the product answers questions in the company's subject area, and how the same question behaves on the other platforms.
The study retains this detail in its tables, rather than treating the three tools as one channel. For online acquisition, separating them recognises that a position achieved as a source in one response does not by itself describe the brand's presence across every AI search experience.
The institutional component of the references
Part of the platform difference appears in the kind of local source cited. Solis groups institutional references through domain patterns identifying government, public-service, academic and financial organisations. Her Spanish examples include Banco de España, on bde.es, and the securities regulator CNMV, on cnmv.es.
In ChatGPT, institutions represent about 46% of the Spanish local citation weight in the study. The table puts them at 11.3 percentage points of all citations, alongside 13.4 points from other .es sources. These are two expressions of the same composition: a share within the local group and a contribution to the overall total have different denominators.
AI Mode shows a different distribution, with 1.8 institutional points and 20.8 from other local sources. Gemini records 0.5 and 16.1 points, respectively. The non-institutional group includes retailers, publishers, services and other organisations. Two platforms with similar local shares can therefore build them from different types of page.
For a company, the distinction helps interpret the space in which it is trying to appear. A regulatory reference and a commercial page may answer different needs within the same query. Recognising each source's function provides more information than grouping every .es domain as though it represented one class of competitor.
Banco de España and CNMV provide concrete examples of this reading in finance. Their presence forms part of the information context the platforms use for the Spanish market. The study shows that localisation can be expressed through recognisable institutions, as well as retailers or content published by brands.
Google and YouTube change AI Mode's position
The comparison includes an adjustment that changes Spain's first place. The domains google.com and youtube.com account for 19.3% of AI Mode's citations in the analysis. Their presence is smaller in the other platforms: 1.0% in ChatGPT and 1.7% in Gemini.
When Solis removes those two domains from the comparison base, AI Mode's .es share moves from 22.6% to 28.1% . ChatGPT stands at 25.0% and Gemini at 16.8%. The ordering of AI Mode and ChatGPT reverses when the remaining source set is compared.
The adjustment changes the measure's denominator. It does not describe an improvement achieved by a brand or a platform's development over time. It answers a different question: how much weight do .es domains have among references remaining after Google and YouTube are excluded? The initial and adjusted rankings provide two views of the same material.
Excluding Google and YouTube changes the comparison
Each platform’s .es share in Spain with all references and after excluding google.com and youtube.com, as published by Aleyda Solis.
Tap or hover for details. Select a series above to show or hide it.
Collapse notes
The exclusion changes the denominator. The two series are alternative readings of the same material, rather than a time trend or performance improvement.
Explore the data table
| Category | All citations | Excluding Google and YouTube |
|---|---|---|
| ChatGPT | 24.8% | 25.0% |
| AI Mode | 22.6% | 28.1% |
| Gemini | 16.6% | 16.8% |
This reading decision has a specific use in competitive analysis. If a team examines sources outside those two domains appearing around its questions, the setting it finds in AI Mode differs from the initial share's suggestion. The research makes that effect visible and prevents the first percentage from becoming a complete explanation of the channel.
The phenomenon also appears in the United Kingdom: AI Mode reaches a 43.0% local share after the exclusion and overtakes the other platforms. Brazil retains a different order, with Gemini ahead. The presence of Google and YouTube explains part of the comparison, but does not create a universal ranking across markets.
The same domain can supply different pages by market
Solis also compares each country's sources with those the same platform uses in the United States. She distinguishes domain overlap from page overlap, identified through the normalised URL. The first measure shows that the same site appears; the second identifies a match for the particular page.
In Spain, ChatGPT shows 51.2% overlap with its US domains and 23.5% page overlap. AI Mode records 36.5% and 2.0%, respectively; Gemini records 32.0% and 17.4%. Reuse of sites coexists with a more differentiated selection of the pages being cited.
The distinction matters for a company publishing across countries under the same brand. A domain can appear in two markets through different content. The same site name does not mean the same page is answering both audiences' questions. URL detail provides a content-level reading that the domain measure does not supply.
In the research, this separation supports examining the localisation of pages as well as the prominence of the site hosting them. A generic domain can offer material adapted for Spain and still be classified as non-local because of its extension. Looking at the page reveals that content within the comparison.
The difference between levels appears in all three platforms, with varying intensity. AI Mode has Spain's smallest page-overlap percentage with the United States. ChatGPT shares more domains, while Gemini occupies another position in both measures. These are signals about source selection, rather than a single scale of which platform understands each country best.
Sharing a site does not mean citing the same page
Domain and page overlap between responses for Spain and the same platform’s US responses, according to Aleyda Solis. Pages are compared through normalised URLs.
Tap or hover for details. Select a series above to show or hide it.
Collapse notes
Each measure retains its comparison unit: site or normalised URL. These are not additive parts, shared users or cross-country traffic.
Explore the data table
| Category | Domain overlap | Page overlap |
|---|---|---|
| ChatGPT | 51.2% | 23.5% |
| AI Mode | 36.5% | 2.0% |
| Gemini | 32.0% | 17.4% |
International strategy requires a local reading
The country comparison reinforces that conclusion. The report publishes local-domain shares of 41.4% in Brazil, 38.3% in the United Kingdom and 24.0% in Spain. In Brazil, Gemini reaches 51.7%, while in Spain it records 16.6%. The same platform offers very different source-selection profiles by market.
AI Mode varies less in its initial share across the three countries, from Spain's 22.6% to Brazil's 27.5%. ChatGPT leads that share in Spain and the United Kingdom. The findings show that the country, topic and platform combination supplies a more useful description than a general label of local or global tool.
The working priorities Solis draws from the research follow that sequence. She first identifies sources cited for important questions in each market, then examines what information those pages provide and whom they serve. From that reading, she proposes prioritising relevant local content and presence on third-party sites appearing in front of the audience.
The process connects source research with a particular acquisition task: understanding the material forming the answer when a user looks for information. In Spain, the differences between finance, beauty, travel and software show why the task needs the context of the business and the product being explained.
The study does not propose changing domains as a universal shortcut. Its comparison brings the relevance of localisation and actual page selection to the foreground. For Spanish brands, the practical result is a more precise map of questions and sources: visibility can concentrate in familiar sites while still requiring different content for each market.
Who's who and how to read the figures
Aleyda Solis is the SEO consultant who conducts and publishes the analysis on her website. Orainti, S.L., identified as the responsible business in that website's privacy policy, is a Spanish company. ChatGPT, Gemini and AI Mode are the three AI response experiences compared; AI Mode is Google's search mode, while Gemini is its other conversational experience.
A ccTLD is a country-associated top-level domain, such as .es. In this study it serves as a geographic indicator, rather than proof of a company's nationality. A generic domain such as .com can publish content targeting Spain. A citation is the reference included by the platform in its answer, distinct from a click or purchase.
Domain overlap identifies the same site across two markets; page overlap compares normalised URLs. Excluding Google and YouTube recalculates the share on another base. The tables belong to research published on 8 October and updated on 10 October. Those dates identify the publication and revision, without turning them into a schedule for collecting the responses.
In this article
Marian Vasile
Marian Vasile is a MarTech and growth professional. His work spans product marketing, SEO, paid advertising and deep online tracking, including behavioral analytics and attribution.