Where Does AI Recommend You Live?
AI Visibility of Residential Developments in Bratislava
Introduction
As more and more people search with AI, brand visibility in language models is becoming increasingly important. According to a report by Demandsage, ChatGPT already has 900 million monthly active users. Search Logistics also reports that as many as 38% of people regularly use AI as a search tool and 37% of buyers start their research with AI. Potential customers also turn to AI on important life matters such as mental health or housing. And housing is the focus of today’s analysis, in which we looked at which residential developments AI recommends in Bratislava.
The data presented here comes from our own research and a prompt set we built systematically to cover queries about the number of rooms and market segment. The prompts asked about neither price nor a specific location. The research design took into account that language models work on the basis of probability and may not always give the same answer to the same question. To avoid this probabilistic bias, we ran the entire prompt set in several rounds. The final dataset on which the whole analysis is based includes a total of 280 results. Further proof of how inconsistent language model answers can be is that several projects appeared in our results under different names, such as Tri veže and III veže. In these cases, we combined the results under a single project. We tested the prompt set on Claude and ChatGPT. ChatGPT is the most popular AI tool in Slovakia, and Claude has long led quality comparisons. We also included Google AI Overviews, which are more closely tied to traditional Google search.
When “Being There” Isn't Enough
The core metric is the so-called Share of Model – the share of all answers in which a given AI model mentions you. This metric is the basic measure of whether AI recommends a particular project at all. The chart below summarizes the projects with the highest average visibility across Claude and ChatGPT. We included only those that passed the 20% threshold.
The clear winner is Slnečnice, which was cited in more than three quarters of answers on average, scoring 76.1%. Although no other project managed to reach visibility above 50%, Bory (41.5%) in second place and Eurovea (36%) in third also scored high. Slnečnice’s score therefore acts as a statistical outlier that skews the overall results in the dataset. Below, we break down the overall results for these projects into results for Claude and ChatGPT.
The key finding from this chart is how differently individual projects feature in each model’s answers. For the outright winner, the gap is negligible, but for Eurovea, Zwirn, Downtown Yards, and SKY PARK, it’s enormous. In practice, this means that to improve visibility, you need to focus on the sources each model draws on.
We also see differences in which projects each model recommends for different housing segments. The table summarizes the top five recommended projects in each segment. Where projects had the same score and therefore shared a place, all of them are listed. To avoid unnecessary noise in the data, we only included projects with more than one mention – that is, at least two.
| SEGMENT | CLAUDE | CHATGPT |
|---|---|---|
| Premium |
|
|
| Starter |
|
|
| Investment |
|
|
| Family |
|
|
What’s interesting about the results is that the models always agreed at least to some extent. Slnečnice was recommended in three of the four segments. Living in Bory is similarly popular, although unlike Slnečnice, the project isn’t always recommended by both models.
We can also translate the project results into results for the developers themselves. Since our analysis focused on projects, we derived the developer results as the sum of the results for their projects. In terms of overall visibility, Cresco Real Estate comes out on top with almost 82%. Developers Lucron (58.5%) and Penta Real Estate (52.2%) also passed the 50% mark, completing the podium.
Where Do Google AI Overviews Come In?
Our research also covered Google AI Overviews. We tested AI Overviews on both full-text prompts and keywords. Google AI Overviews appeared in 61 of 80 full-prompt searches and in 2 of 80 keyword searches. In total, the feature was triggered 63 times, with the overwhelming majority triggered by longer queries.
We split the Google AI Overviews results into mentioned brands (entities) and cited websites. The results are based on the share of searches in which Google AI Overviews appeared. In other words, much like in an election, the results are calculated from “votes cast,” not from “all eligible voters.”
Slnečnice once again tops the ranking by a wide margin. Compared with the other successful brands, however, its 76.2% score can be seen as a statistical outlier. AI Overviews also frequently mention SKY PARK (30.2%) and Čerešne (22.2%), which round out the top three. Interestingly, only two development companies made it past the 10% mark: YIT Slovakia (17.5%) and ITB Development (11.1%).
As part of the analysis, it’s also useful to know where Google AI Overviews get their information.
The results show that the go-to source is the portal bratislavaliving (69.8%), in first place just ahead of nehnutelnosti.sk (63.5%). The project website slnecnice.sk (58.7%) rounds out the top three most-cited websites, while zoznamrealit.sk also passed the 50% mark with 54%.
Conclusion
AI now complements classic Google search for many kinds of questions. Brand visibility in AI chatbot answers is therefore key to commercial success, and understanding why one model cites your brand while another doesn’t is becoming valuable expertise. According to the Search Logistics report cited above, as many as 70% of companies achieve a better return on investment by focusing on AI visibility.
Do you want to receive our data and analyses and be part of the community?
