Real estate + AI search

The increasing adoption of AI tools is changing how brands achieve digital visibility.

Unlike traditional SEO, which focuses on information retrieval and ranking within the top presented results for clicks, AI chatbot responses are single-output, conversational, and brand-selective. This offers fewer exposure opportunities—but greater influence—for well-optimized brands. The nature of AI search also has the benefit of moving users further down the funnel, faster.

This has led to the emergence of generative engine optimization (GEO), which focuses on how brands are interpreted, mentioned, and cited by AI. But how do different GEO strategies play out in practice? And if there’s overlap between GEO and SEO, how strongly do Google rankings translate to LLM visibility? I did an analysis of real estate websites to learn more about GEO and SEO in this particular sector. Here’s what I found.

The question:

Does top performance in Google’s “traditional” organic results contribute to LLM visibility for real estate websites?

The experiment:

Step 1: Determine the brands that rank highest in Google search

First, I identified the top organic performing US real estate websites around the category “Homes for sale.” I used this category rather than keywords because of the probabilistic nature and “reasoning” of LLMs.

These were the top performers in Google search: 

• Zillow.com

• Craigslist.org

• Realtor.com

• Redfin.com

• Homes.com

• Trulia.com

Zillow dominates traditional search compared to brokers such as Keller Williams, RE/MAX, and Coldwell Banker.

Step 2: Analyze LLMs and the different factors that contribute to visibility

I used Evertune AI to analyze these models:

• ChatGPT

• Gemini

• Meta

• Claude 

• DeepSeek

First, I created a Word Association report. This works by prompting AI models to generate multiple reviews for a brand and its product category, representing a typical conversation a customer would be having with an LLM about the brand. 

Then, Evertune AI analyzed these reviews to identify frequently used keywords and assigned two scores to each:

• Association Score. This score, ranging from 0 to 100, indicates how frequently a keyword appears in the reviews.

• Sentiment Score. This score, from -100 (negative) to +100 (positive), reflects the sentiment associated with the keyword.

Using Evertune AI, I generated a word cloud in which the size of each word corresponds to its Association Score, and its color is determined by its Sentiment Score.

I was also able to assess brand strength using Evertune AI’s AI Brand Index. To measure brand strength, Evertune AI prompts LLMs thousands of times to see which brands surface for questions like, “what are the brands you think of first in X?” Only strong brands will surface here.

I also looked at a model's likelihood of recommending a brand unprompted, based on a specific consumer preference, like homes in a certain location or condition. Essentially, which brands show up when you search for a newly built condo near the beach in Miami?

Ultimately, I looked at these four factors—organic rankings, brand sentiment, brand strength, and consumer preferences—to see how each affected a brand’s visibility in LLMs.

The findings:

Organic rankings 

Lots of people think success in Google search directly translates to success in LLMs. This research shows that it doesn’t. Incumbent players like Craigslist that continue to perform well in traditional organic visibility don’t automatically do well in AI synthesis. In fact, Craigslist's probability of being mentioned in AI outputs was zero, across all models in Evertune AI’s Brand Index. 

If you recall from the chart above, Zillow performed the best in Google search. But not in LLMs. Brokers such as RE/MAX, which had an 81% chance of being mentioned, performed better than marketplaces across the models. Zillow had a 31% chance of being recommended.

Brokers were 175% more likely to be recommended than marketplaces. I concluded this by taking the mean average of "Brand Score" for brokers (57.3) vs. the mean average of marketplaces (20.8), calculated the absolute difference (36.5) and expressed this difference as a percentage of the marketplace score (the baseline for comparison).

Mean overall "Brand Score" was across AI models: GPT 4., Gemini 2.0, Llama 4, Claude Sonnet 4, and Deep Seek v3.

This is something we’ve observed across other industries as well, and not exclusively limited to real estate. My theory is that LLMs are designed to provide concise, direct answers, aiming for a zero-click experience where the user doesn't need to leave the AI interface. If a user asks for a recommendation, providing a specific brand name and product can feel like a more complete and helpful answer than simply listing a marketplace, where the intent is to search more.

Brand sentiment

Craigslist performed the worst on negative sentiment and was the lowest recommended brand across all models analyzed. Particularly salient negative terms included: “lack of regulation,” “outdated listings,” “fraudulent listings,” and “lack of verification.”

Redfin performed the best with sentiment and in particular were salient around the terms “user-friendly interface,” “high quality photos,” and “virtual tours.” Craiglist’s sentiment score was zero compared to Redfin’s 71/100.

UGC (such as Reddit and Facebook) and review sites (such as Yelp and Yellow Pages) seem to be major contributing factors to sentiment.

Data from Similarweb shows that there is some correlation between the percentage of referral traffic from LLMs and the sentiment score seen on these tools. For instance, we see that Redfin has the highest sentiment score and also has the highest referral percentage from AI search. On the other end of the scale, Craigslist has the lowest sentiment score and also the lowest percentage of referral traffic from AI.

Brand strength

The AI Brand Index, developed by Evertune AI, measures how likely an AI model is to recommend a brand without being prompted by the user's question. This index serves as a high-level brand tracking metric, offering insights into AI's unaided awareness of various brands. 

It works by posing thousands of questions to large language models (LLMs) across six brand vectors, mimicking typical survey questions a human researcher would use to gauge unaided brand awareness within a product category. The goal is to assess how often an AI model recommends a specific brand and its perception compared to competitors in the same category.

Keller Williams, RE/MAX, Coldwell Banker, Compass, and Redfin were most likely to be mentioned in AI outputs, even though they weren't in our original seed of sites based on US organic visibility today. All three scored over 60% probability of being mentioned in AI outputs, compared to Zillow at 33%.

The data consistently shows that brands with higher AI Brand Scores also have higher visibility scores across the various Consumer Preferences (see below). This indicates that improving AI visibility is strongly linked to building a strong overall brand presence and being "top of mind."

Consumer preferences

"Consumer preferences" from Evertune AI measures a model's likelihood of drawing unaided attention to a brand within a category based on a specific consumer preference, such as "fit" or "comfort" for shoes.

To achieve this, Evertune AI identifies the most important attributes AI models associate with a given product category. Subsequently, the models are sampled to determine which brand is most frequently recommended for each attribute. This data is then used to calculate an "AI Attention Score" for each product attribute, where a score of 43 indicates a 43% probability of the AI recommending a brand to a user for that particular preference.

These are the attributes the models think are most important in the category of “homes for sale:”

• Location

• Price

• Size

• School district

• Condition

• Neighborhood

• Safety

• Proximity to amenities

• Neighborhood safety 

• Property taxes

Brands that want to improve their visibility within the models need to give attention to these particular facets, which vary according to each category.

"Location," "School district," "Condition," "Neighborhood," and "Safety" often have high AI Brand Scores and Visibility Scores for top-ranked brands like Keller Williams, RE/MAX, and Coldwell Banker.

Zillow should focus on improving their visibility around key facets such as "Neighborhood"

with earned and onsite media focused on comprehensive neighborhood insights. They would also do well by developing an engaged user community to foster authentic conversations and firsthand experiences about neighborhoods.

"Property Taxes" are a particular strength for Zillow, largely via their Zestimate tool and subsequent earned media which references property taxes, as well as comprehensive onsite content covering this area.

Redfin, averaging across all models, seems to do best on answering these model preferences. However, their lower brand strength led to less overall visibility in AI outputs.

Summary

Historical organic search performance doesn't translate into the AI synthesis of today. Models respond based on what they think are important consumer preferences. Brand influence and sentiment are far more important than legacy SEO factors. These findings are consistent with a similar report I conducted on the UK real estate market.

From my findings, looking at hundreds of reports now across varying industries, the order of importance would be:

• Brand strength

• Brand sentiment

• Consumer preferences

• Organic rankings (yes, this is at the bottom of the list)

This means brands need to adapt their strategies to ensure visibility in AI search, focusing on “category” brand strength and sentiment and how topics are covered holistically, both on-site and off-site. Brands with deep content coverage on specific topics may be mentioned more frequently by AI chatbots, even over larger, more authoritative domains with less specific content. 

It's also important to clearly communicate brand benefits to offset potential negative sentiment, such as higher costs, and to monitor forums and review sites to identify and address negative connotations about your brand.

Overall, AI presents a new and critical touchpoint for brands to engage with their audience and demonstrate their unique value. Brands that prioritize strong storytelling, reputation management and digital PR to build authority and trust will be well-positioned for discoverability via AI search.