Over the past 18 months, we've been running what we believe is the most comprehensive cross-engine brand recommendation study in the GEO industry.
50,000 AI-generated responses. Four major engines (ChatGPT, Claude, Gemini, Perplexity). 25 industries. 2,500 unique queries. Tracked monthly from mid-2025 through January 2026.
The goal was simple: figure out what actually determines which brands AI recommends. Not theories. Not case studies of one company. Hard data across thousands of data points.
Here's what we found.
Methodology
Before the findings, the methodology matters. Here's exactly what we did:
Query design: 100 queries per industry, covering four intent types:
- Direct recommendations ("What's the best X for Y?")
- Comparative queries ("X vs Y vs Z")
- Problem-solving queries ("How do I solve X?")
- Research queries ("What should I consider when choosing X?")
Data collection: Each query was run 4x per month (weekly) across all four engines. Responses were parsed for brand mentions, positioning (1st mention, 2nd, etc.), sentiment, and citation sources.
Analysis: We used statistical modeling to identify which brand attributes correlated most strongly with recommendation frequency, controlling for brand size, industry, and engine.
The Five Factors
Factor 1: Source Diversity Score (r = 0.73)
The single strongest predictor of AI recommendation is how many different types of sources mention your brand positively.
Not just how many sources—how many types. A brand mentioned in 50 blog posts has a lower Source Diversity Score than a brand mentioned in 10 blog posts + 5 Reddit threads + 3 YouTube videos + 2 industry reports + 1 Wikipedia mention.
The numbers:
- Brands in the top quartile of Source Diversity were recommended 4.2x more often than bottom quartile brands
- The threshold effect was striking: brands mentioned across 5+ source types saw a dramatic jump in recommendation frequency
- Below 3 source types, brands were recommended less than 8% of the time regardless of other factors
Why this matters: AI models are trained to synthesize information across diverse sources. A brand that appears consistently across multiple source types gets stronger signal reinforcement. One-dimensional presence (even if deep) is less effective than broad, diversified presence.
Factor 2: Positioning Clarity Index (r = 0.61)
We developed a Positioning Clarity Index by measuring how consistently AI models could categorize a brand across repeated queries.
If ChatGPT calls you a "project management tool" on Monday and a "collaboration platform" on Thursday, your Positioning Clarity is low. If all four engines consistently describe you the same way, it's high.
The numbers:
- Brands with high clarity were recommended 2.8x more often for category-specific queries
- 41% of brands had significant positioning inconsistency across engines
- The most common problem: different messaging on the company website vs. third-party profiles vs. press coverage
Why this matters: When a user asks "What's the best project management tool?", the AI needs to match brands to that category. If your positioning is ambiguous, the AI can't confidently categorize you, so it defaults to brands with clearer positioning.
Factor 3: Recency-Weighted Authority (r = 0.58)
Not all mentions are equal. We found that recency dramatically amplifies authority signals.
A brand mentioned in a 2025 industry report gets weighted significantly more than one mentioned in a 2023 report—even if the 2023 report is from a more authoritative source.
The numbers:
- Mentions from the past 6 months had 3.1x the impact on recommendations compared to mentions 12-18 months old
- This effect was strongest for Perplexity (4.5x recency multiplier) and weakest for ChatGPT (2.2x)
- Brands that stopped producing citeable content saw their recommendation frequency drop an average of 15% per quarter
Why this matters: AI models—especially those with real-time search capabilities like Perplexity—heavily weight fresh information. A strong presence that isn't maintained will erode. GEO is not a one-time project.
Factor 4: Sentiment Consistency (r = 0.52)
We measured not just whether sentiment was positive, but whether it was consistently positive across sources.
A brand with 80% positive mentions across all source types performed better than a brand with 95% positive mentions from company-controlled sources but mixed sentiment on community platforms.
The numbers:
- The positive-sentiment threshold for strong recommendations was 70%+ across all source types
- Negative Reddit sentiment had the single largest negative impact on Perplexity recommendations (-34% recommendation frequency)
- Interestingly, brands with some authentic negative mentions alongside many positives performed better than brands with 100% positive (but potentially curated) sentiment
Why this matters: AI models appear to factor in sentiment authenticity. A brand that has 100% positive mentions everywhere looks suspicious. A brand with overwhelmingly positive mentions plus some genuine criticism (that's addressed transparently) looks trustworthy.
Factor 5: Entity Graph Completeness (r = 0.47)
This factor measures how well-defined your brand is as an entity in AI's knowledge—essentially, how much structured information exists about your brand across the web.
Components include:
- Google Knowledge Graph presence
- Consistent NAP (Name, Address, Phone) data
- Schema.org markup on your website
- Wikipedia/Wikidata presence
- Structured review data (aggregate ratings, review counts)
The numbers:
- Brands with complete entity graphs were recommended 1.9x more often
- Schema markup alone improved factual accuracy by 38% (confirming earlier research)
- Knowledge Graph presence had the strongest effect on ChatGPT, less on Perplexity
Why this matters: A well-defined entity gives AI confidence. When the model can verify facts about your brand through structured data, it's more willing to recommend you.
The Interaction Effects
Individual factors tell part of the story. But the real insight is in how they interact.
Source Diversity × Positioning Clarity: Brands strong in both were recommended 6.7x more than brands weak in both. These factors are multiplicative, not additive.
Recency × Sentiment: Fresh positive mentions were worth 5x more than old positive mentions. Fresh negative mentions were worth -3x. Recency amplifies whatever sentiment exists.
Entity Completeness × Source Diversity: Having a strong entity graph amplified the impact of diverse mentions by 40%. The AI "trusts" mentions more when it has a clear entity to attach them to.
What This Means For Your Strategy
Based on the data, here's the optimal resource allocation for GEO efforts:
Source Diversity (30% of effort): Prioritize getting mentioned across multiple source types. If you're only on blogs, expand to Reddit, YouTube, industry reports, podcasts.
Positioning Clarity (25%): Audit every place your brand appears online. Unify your messaging. Make it impossible for AI to miscategorize you.
Recency (20%): Establish a cadence for creating citeable content. Monthly at minimum. Quarterly content that's worth citing is better than weekly content that nobody references.
Sentiment (15%): Monitor community sentiment actively. Address negative feedback transparently. Don't try to suppress criticism—respond to it.
Entity Completeness (10%): Implement Schema markup, maintain your Knowledge Graph presence, keep structured data current.
The 80/20 of Brand Recommendations
If I had to boil 50,000 data points into one sentence, it would be this:
AI recommends brands that are clearly defined, widely discussed across diverse sources, recently validated, and consistently well-regarded.
That's not groundbreaking on the surface. But the data behind it quantifies exactly how much each factor matters—and reveals that most brands are investing their GEO efforts in exactly the wrong proportions.
The full dataset is informing everything we build at GeoBuddy. This study will continue running through 2026, and we'll publish updated findings as patterns evolve.
Originally published on GeoBuddy Blog.
Is your brand visible in AI answers? ChatGPT, Claude, Gemini & Perplexity are shaping how people discover products. Check your brand's AI visibility for free — 3 free checks, no signup required.
Top comments (0)