How Local Businesses Can Win in AI Search: Lessons From 120,000 LLM Recommendations

Understanding AI Search Personalities

A new study by local marketing platform Uberall has shed light on how generative engine optimization (GEO) operates across different artificial intelligence models. While search engine optimization historically focused on understanding Google's algorithms, local brands now face the task of optimizing for multiple AI platforms, each exhibiting its own distinct "personality" when recommending local businesses.

The study, conducted by Uberall's internal GEO analyst Katya Shishchenko, tracked over 120,000 AI mentions across 3,793 business locations in major US cities, including Chicago and New York. The researched locations spanned several verticals: dentists, restaurants, grocery stores, hotels, and banks.

According to the findings, the major AI engines vary significantly in their approach:

• Claude: This model behaves conservatively. It prioritizes local and community-focused businesses, but it largely avoids naming specific healthcare providers, likely to prevent giving medical advice.
• Gemini: Characterized as the most diverse, Gemini pulls live Google Maps data to surface results. This integration allows it to recommend eight times more unique restaurants than ChatGPT in identical scenarios.
• ChatGPT: Known for being consensus-driven and concentrated, this model generates a short, consistent shortlist, but it also records the highest rate of hallucinations among the tested models.
• Grok: Behaving like a long-form food journalist, Grok frequently references chef qualifications, cooking techniques, historical context, and Instagram content. It often highlights "must-try dishes" and specific chef names.
• Perplexity: Serving as the "easiest" platform to win on for businesses with an existing web presence, Perplexity conducts live web searches, provides citations for its sources, and outputs the highest volume of mentions per run.

Decoding the "BARS" Framework for AI Visibility

Despite their individual behavioral differences, the analyzed AI models consistently reward similar foundational signals. Uberall categorizes these underlying elements under the acronym BARS:

• Business data
• Authority
• Review
• Social

The Critical Role of Basic Business Data

The Uberall research emphasizes that basic business data completeness—specifically GBP (Google Business Profile) details and business age—serves as the primary gatekeeper. These elements determine whether a local business is deemed eligible to be mentioned by an AI model at all, rather than how frequently it appears once it becomes part of the candidate pool.

While business age cannot be changed, marketers can optimize other critical data points that directly impact AI visibility across different industry verticals:

1. GBP Descriptions and Attributes

Providing thorough information dramatically increases an AI's likelihood of recommending a business. For grocery stores, completing the GBP description can trigger a threefold increase in AI mention rates. Meanwhile, in the hotel sector, expanding the listed attributes from a minimal 6–10 up to 31–50 moves the probability of being mentioned from a mere 22% to an overwhelming 94%.

2. Photo Volume

The volume of photos associated with a business profile acts as a powerful driver for several industries. In the restaurant vertical, photo count stands as the single strongest predictor of how often a business is mentioned, with the most-mentioned eateries averaging three times more photos than their competitors. For dental practices, photo count is unique in predicting both whether a clinic is mentioned and how frequently. It also ranks as one of the three most powerful predictors of AI mentions for banking institutions.

3. Location Scale

Having multiple locations boosts overall brand visibility within AI models. In both the dental and grocery sectors, businesses that reach a scaled threshold of multiple locations achieve a 100% mention rate.

Ultimately, optimizing for AI local search does not require exotic strategies. Instead, it relies on executing foundational local data management intentionally and at scale.