Your Brand Has a Different Reputation on Every AI Platform

Sill's monitoring pipeline queries four AI platforms daily through their actual chat interfaces: ChatGPT, Gemini, Google AI Overviews, and Perplexity. Each platform receives the same set of purchase-intent queries for each brand, with web search enabled and citations extracted. This analysis covers 7,442 individual query-response pairs across 139 brands spanning 19 industries.

Every response is analyzed for Share of Voice (SOV): whether the target brand is mentioned, how prominently it appears relative to competitors, and the sentiment of the mention. Per-platform SOV scores range from 0 (not mentioned) to 100 (dominant recommendation). The aggregate SOV is the average across platforms.

The query distribution is balanced: ChatGPT (25.1%), Gemini (25.2%), Google AI Overviews (25.1%), and Perplexity (24.6%). This balance is by design. Every brand is measured on every platform with the same queries.

Across all 139 brands, mean SOV scores vary by platform. The gap between the most generous platform (Google AI Overviews, 19.8) and the most conservative (Perplexity, 16.0) is 3.8 points at the aggregate level. The real story is in the per-brand variance.

The median SOV is identical across all four platforms (15.0), meaning the "typical" brand experience is similar everywhere. The divergence shows up at the tails: some brands dramatically over-perform on one platform and under-perform on another. A 3.8-point difference in the mean masks brand-level spreads of 30, 40, and even 50 points.

For each brand, we calculated the spread: the difference between the highest and lowest platform SOV score. The results are stark.

• Average spread per brand: 11.7 points. On a 0-100 scale, this means the typical brand's best platform scores nearly 12 points higher than its worst.
• 61% of brands have a spread greater than 5 points. The majority of brands would get a meaningfully different picture of their AI visibility depending on which platform they checked.
• 55% have a spread greater than 10 points. More than half of all brands have at least one platform where they are substantially more visible than another.
• Maximum observed spread: 50 points. One brand scored 50.0 on Google AI Overviews and 0.0 on ChatGPT. Same queries, same day.

If a marketer checks their brand's AI visibility on only one platform, they have roughly a coin-flip chance of getting a score that differs by 10+ points from their score on a different platform. Single-platform monitoring is a partial measurement. Multi-platform monitoring is the minimum.

SOV divergence is not limited to the target brand. Competitor rankings shift across platforms. In our monitoring data, 76 competitors showed rank divergence of 2 or more positions. The average rank spread was 6.1 positions, with a maximum of 16 positions.

Consider a real example from the data. In the fitness wearables category, Headspace ranked #3 on Perplexity (SOV 27.3), #6 on Google AI Overviews (SOV 12.5), and #19 on ChatGPT (SOV 10.6). A brand competing with Headspace would draw entirely different competitive conclusions depending on which platform they analyzed.

Apple provides another example. In one product category, Apple ranked #2 on Perplexity (SOV 35.8), #4 on both Gemini and Google AI Overviews, but #15 on ChatGPT (SOV 10.6). In a different category, Apple ranked #7 on ChatGPT (SOV 23.8), #20 on Gemini (SOV 2.3), and #5 on Google AI Overviews (SOV 19.2). Same brand, same timeframe, different platforms, different competitive positions.

The competitive landscape is not fixed. It is platform-dependent. A brand that appears dominant on ChatGPT may be barely visible on Perplexity, and vice versa. Competitive analysis that checks only one platform will identify the wrong threats and the wrong opportunities.

The divergence is not random. Each platform uses a different retrieval pipeline, draws on different source pools, and applies different ranking logic to the content it finds. CMU's AutoGEO research (Wu et al., 2025) found that engine-specific optimization rules consistently outperform generic strategies, achieving 35.99% average improvement. The reason is that each engine rewards different content signals.

ChatGPT uses Bing-based web search, which favors established domains and pages with strong backlink profiles. Perplexity runs aggressive multi-source retrieval and leans heavily on Reddit, forums, and user-generated content. Gemini draws on Google Search infrastructure and tends to favor structured, authoritative lists. Google AI Overviews pull directly from organic search results and overlap significantly with traditional SEO rankings.

These architectural differences explain why a brand with strong Reddit presence over-performs on Perplexity, while a brand with strong traditional SEO performs best on Google AI Overviews. The brand did not change. The retrieval lens changed.

The SOV divergence is mirrored in citation behavior. Across 23,710 URLs cited by AI platforms in our pipeline, 91.6% were cited by exactly one platform. Only 8.4% of URLs were cited by two or more platforms. Just 0.1% were cited by all four.

This means each platform is reading a nearly independent set of sources when forming its answer. A page that drives your visibility on Perplexity is almost certainly not the same page driving visibility on ChatGPT. The anatomy of what gets cited varies by platform. Citation data from one platform tells you almost nothing about your citation footprint on another.

This finding aligns with the Ahrefs LLM visibility study (75,000 brands): only 11% of domains cited by ChatGPT also appear in Perplexity citations. Our page-level data confirms and extends this: the overlap is not just low at the domain level, it is low at the individual URL level.

Beyond visibility scores, the way platforms talk about brands differs. We classified sentiment across 3,371 AI responses into four categories: very positive, positive, neutral, and negative.

The most notable pattern: Perplexity is significantly more neutral than other platforms. 44.5% of Perplexity responses carry neutral sentiment, compared to 34.3% for Gemini and 37.6% for ChatGPT. Perplexity is also far less likely to express strong enthusiasm (1.7% very positive vs. 7.1% for Gemini).

Gemini, on the other hand, has the highest negative sentiment rate (2.5%) and the highest very positive rate (7.1%). Gemini tends to take stronger positions. Google AI Overviews is the most conservative, with the lowest negative rate (0.9%) and moderate positive framing.

For a brand monitoring its AI perception, these differences matter. A brand might be described enthusiastically on Gemini and flatly on Perplexity. If you are only monitoring one platform, you are only seeing one version of how AI talks about you.

SparkToro's research (2026) demonstrated that AI recommendation lists repeat less than 1% of the time across identical prompts. Our data adds a second dimension: even when a brand does appear, its relative standing varies dramatically by platform. The instability is not just temporal. It is structural.

The practical implication is that "AI visibility" is not one metric. It is four (at minimum). A brand with high ChatGPT SOV and zero Perplexity SOV does not have "moderate AI visibility." It has a platform-specific distribution that requires platform-specific optimization. The foundational GEO paper (Aggarwal et al., KDD 2024) showed that different optimization strategies work on different engines. Our data confirms that brands need to measure each platform independently to know where to focus.

Three actions follow from this data:

1. Monitor all platforms independently. An aggregate SOV score obscures platform-specific strengths and weaknesses. Track ChatGPT, Gemini, Google AI Overviews, and Perplexity separately.
2. Identify your weakest platform. With a median spread of 12.5 points, most brands have at least one platform where they significantly underperform. That platform is where the highest-leverage optimization opportunity exists.
3. Track competitor rankings per platform. Your #1 competitor on ChatGPT may be a different company than your #1 competitor on Perplexity. Competitive strategy should account for this.

Sill's monitoring dashboard tracks SOV, sentiment, competitor rankings, and citation provenance independently across every platform, updated daily. Each metric can be filtered by platform, product category, buyer persona, and geographic location.

Because Sill monitors through the actual chat interfaces, not APIs, the data reflects what buyers see when they ask each platform for a recommendation. When your team makes a GEO optimization, Sill shows whether it moved the needle on ChatGPT, Perplexity, or both. The Semantic Map adds a perception layer, showing how each platform positions your brand on dimensions like price, innovation, or reliability.

Platform divergence is not a bug in AI search. It is a structural feature of how these systems work. The brands that understand and measure it will optimize more effectively than those treating AI as a monolith.