Across 212 AI visibility audits, ReSO found that AI engines rarely agree on which brands to recommend. Only 5.4% of prompts returned the same #1-ranked brand across ChatGPT, Perplexity, and Google AI Overviews, while 77.3% of brands appeared on just one of the three engines.
The differences go beyond rankings. Even when the same brand appeared across multiple engines, each engine showed its own preferences in the brands it surfaced and the sources it trusted. AI visibility is not a single leaderboard or a single score. It is three distinct competitive landscapes that brands need to measure and optimize independently.
Key findings
- AI engines rarely agree on the same recommendations: Only 5.4% of prompts returned the same #1-ranked brand across ChatGPT, Perplexity, and Google AI Overviews.
- Most brands are visible on just one engine: 77.3% of brands appeared on only one AI engine, showing that AI visibility is highly fragmented.
- Engine divergence happens at three levels: AI engines differ in the brands they surface, the rankings they assign, and the sources they cite to support their answers.
- A single AI visibility score hides important differences: Measuring visibility separately for each engine provides a clearer picture than averaging performance across all three.
- Each engine relies on different authority signals: ChatGPT, Perplexity, and Google AI Overviews draw from different source classes, creating unique visibility opportunities for every engine.
- AI visibility strategies should be engine-specific: Prioritizing the engines your buyers actually use leads to more effective optimization than treating AI search as a single channel.
The three layers of engine divergence
Engine divergence surfaces at three levels of the visibility stack, and each level is more consequential than the last.
Layer 1: Engines surface different sets of brands
Across 65,976 unique canonical brands captured across our audit corpus, only 8.1% appear on all three engines. 77.3% appear on exactly one engine. Another 14.6% appear on exactly two. Put more concretely: the ChatGPT-only pool alone (20,876 brands) is larger than the entire on-all-three set (5,332 brands). The pattern is not a marginal recall difference between three broadly similar systems. It is three overlapping but materially different brand universes, and whichever engine a brand’s visibility audit runs against determines most of what it sees.
Layer 2: Engines rarely agree on the top-ranked brand
Where all three engines returned an answer to the same prompt (6,348 prompts in our corpus), they named the same brand at position #1 in 340 of them. That is a 5.4% agreement rate. Read the inverse: 94.6% of the time, the “top answer” a buyer sees depends entirely on which engine they consulted. “Ranking first on AI” is not one status. It is three separate statuses that rarely coincide, and a brand that ranks first on ChatGPT for a category-defining query cannot assume the equivalent rank on Google AI Overviews or Perplexity.
Layer 3: Engines cite different sources, and cite them unequally
The divergence goes beyond which brands surface. Of the 5,316 brands that do appear on all three engines, only 69.1% are cited with a balanced distribution across engines. The other 30.9% lean heavily on one engine, meaning even brands present in every engine’s pool are not present equally. Beneath that, the named brand-preference patterns are legible in the data. Google AI Overviews leans on analyst and professional-network brands: Gartner appears with a 91.5% AIO share of its total mentions, LinkedIn 69.4%, and AWS 67.4%. ChatGPT surfaces a broader recall, including newer and category-challenging brands. Perplexity sits between the two with its own preferred cluster.
The source-class divergence sits on top. When engines cite third-party pages to justify a recommendation, they draw on different pools. In our informational citation set, ChatGPT effectively did not cite YouTube; Perplexity and Google AI Overviews together carried all 3,440 YouTube citations recorded. Reddit’s informational citations were 96.5% ChatGPT-attributed. Wikipedia’s informational citations were 99.7% ChatGPT-attributed. Regulatory sources skewed ChatGPT-first heavily. The three engines are not returning different brands only. They are drawing on entirely different source classes to justify those brands.
Why single-number AI visibility metrics miss most of the picture
A single AI visibility score averages your performance across ChatGPT, Perplexity, and Google AI Overviews. While that number may look useful, it can hide significant differences in how your brand performs on each engine.
Take a brand with an average AI visibility of 30%. That score could represent several very different situations:
- 30% visibility across all three engines.
- 60% on ChatGPT, 20% on Google AI Overviews, and 10% on Perplexity.
- 90% on Google AI Overviews and 0% on the other two.
Each scenario requires a different strategy, even though the overall visibility score is identical. The 5.4% top-position agreement rate shows that these differences are the rule rather than the exception.
An overall AI visibility score cannot tell you which engine your buyers use, where your brand is strongest, or which engine needs the most attention. As AI answers capture a growing share of buyer attention, those distinctions become even more important. SparkToro’s 2026 research found that 68% of Google searches in the US end without a click, meaning more buying journeys are being shaped directly on the AI answer surface.
Building a visibility strategy across three divergent engines
If AI engines diverge on brands, rankings, and sources, your visibility strategy should reflect those differences. ReSO’s engine-triangulated visibility framework treats each engine as an independent citation surface, with its own measurement, gap analysis, and optimization priorities.
Component 1: Measure each engine independently
Run the same prompt set across ChatGPT, Perplexity, and Google AI Overviews, then measure each engine separately.
Track:
- Which brands appear on each engine?
- Where your brand ranks on each engine.
- Which third-party domains and source classes each engine cites?
Avoid combining the results into a single visibility score. Per-engine metrics such as share of voice, average position, and citation-source distribution reveal insights that an overall average cannot. Brand safety monitoring also benefits from this approach, since a brand misrepresentation issue on one engine may never appear on another.
Component 2: Diagnose each engine’s visibility gaps
Treat absence on each engine as a different problem.
- ChatGPT gaps often reflect limited Reddit or Wikipedia coverage, two of its dominant informational source classes.
- Google AI Overviews gaps often point to a weaker LinkedIn presence, limited video content, or low visibility across analyst and professional-network sources.
- Perplexity sits between the two, with its own weighting of citation sources.
Identifying which engine your brand struggles with is the first step. The corrective tactics differ from one engine to another.
Component 3: Prioritise based on buyer behaviour
If your buyers primarily use ChatGPT, make ChatGPT visibility the first priority. If they rely more on Google AI Overviews, focus there first. Buyer interviews, usage research, or category behaviour can help determine which engine deserves the greatest attention.
Do not assume that the engine with the largest brand pool is automatically the most important. In our dataset, ChatGPT surfaced 32,020 unique brands, but that does not necessarily reflect where your buyers spend the most time.
How to monitor engine drift after the initial audit
The findings are based on ReSO’s proprietary AI visibility dataset covering 212 brand audits conducted between March and June 2025. Each audit included approximately 35 commercial-intent prompts and 15 informational-intent prompts, with every prompt run across ChatGPT, Perplexity, and Google AI Overviews.
Brand mentions from each engine were captured, normalized to a canonical brand ID, and used to calculate engine-overlap statistics. Top-position agreement was measured across the 6,348 prompts where all three engines returned a first-position brand, allowing us to compare not only which brands appeared but also how consistently the engines ranked them.
A few methodology notes:
- Brand normalization is not perfect. A small number of collision artifacts, most notably an “Austin” cluster, were identified and excluded from the Layer 3 analysis.
- Per-engine citation splits are currently available only for the informational dataset. As a result, all Layer 3 source-class findings apply to informational queries only.
This analysis focuses on how AI engines differ in the brands they recommend, the rankings they produce, and the sources they rely on. The underlying informational-versus-commercial citation regimes are explored separately in the earlier analysis, while the citation snowball effect, where citations compound over time as engines repeatedly reference the same trusted sources, is explored in a companion piece.
How the engine-level data was gathered
The dataset is the same proprietary corpus of AI visibility audits ReSO ran across 212 brands between March and June 2025. Each audit fired approximately 35 brand-inducing prompts and 15 informational prompts against ChatGPT, Perplexity, and Google AI Overviews. Brand mentions per engine were captured and normalized to a canonical brand ID; engine overlap statistics were computed by set operations across the per-engine mention lists.
Top-position agreement was computed on the 6,348 prompts where all three engines returned an answer with a first-position brand named. Two notes on scope: brand normalization is imperfect and a small number of collision artifacts (most notably an “Austin” cluster) were excluded from Layer 3 computations.
Per-engine citation splits at the domain level are available for the informational half of the dataset in this pass; Layer-3 source-class figures apply to informational queries only. Piece one covers the underlying informational versus commercial citation regime separately.
Get your engine-triangulated AI search visibility audit
A single AI visibility score cannot show how your brand performs across ChatGPT, Perplexity, and Google AI. Each engine recommends different brands, ranks them differently, and relies on different sources to support its answers.
ReSO’s AI Visibility Audit measures your visibility separately for each engine. It shows where your brand appears, which competitors are surfaced instead, the source classes each engine relies on, and the opportunities to strengthen your visibility where your buyers are most likely to search.
Every audit includes a per-engine visibility breakdown and a prioritised action plan, helping you focus on the engines that matter most for your category. Request your AI Visibility Audit to see where your brand stands across ChatGPT, Perplexity, and Google AI.
Frequently asked questions
Does the 5.4% top-position agreement mean the engines are wrong?
No. It means the three engines selected different top-ranked brands for the same query. Each engine uses its own retrieval and ranking systems.
Should a brand focus on one engine or all three at once?
It depends on where your buyers search. Prioritize the engines your audience uses most instead of treating every engine equally.
Why does ChatGPT surface so many more unique brands than the other two?
ChatGPT retrieves a broader range of sources, including Reddit and long-tail documentation, resulting in a larger brand pool. A larger pool reflects broader coverage, not necessarily higher-quality recommendations.
Does vertical composition affect engine agreement?
Yes, but only slightly. SaaS shows higher cross-engine agreement than Services, although agreement remains low across every industry.
