Not sure where to start? Take a guided tour.
AI Site Grader
AI Query Scanner
llms.txt GeneratorShould you use one model as a proxy for all AI visibility?
No. With only 43.3% average agreement and 4.0% perfect consensus across 8 models, one model is an unreliable proxy for the wider AI market.
The key numbers.
- Average agreement
- Mean cross-model agreement rate.
- 43.3%
- Perfect agreement
- Only a small share of prompts produce unanimous outcomes.
- 4.0%
- Models analyzed
- OpenAI, Anthropic, Gemini, Grok, Deepseek, Meta, Perplexity, and Google AI Overviews.
- 8
Visibility must be measured across the specific model set relevant to your audience, as inferring market wide performance from a single platform yields inaccurate data.
The evidence.
| Measure | Result |
|---|---|
| Average agreement Mean cross-model agreement rate. | 43.3% |
| Perfect agreement Only a small share of prompts produce unanimous outcomes. | 4.0% |
| Models analyzed OpenAI, Anthropic, Gemini, Grok, Deepseek, Meta, Perplexity, and Google AI Overviews. | 8 |
Population and method
Built from 797,644 valid comparisons across 44,088 reports and 8 models, covering 6,439,133 model responses in the observed window.
Where the evidence stops
Agreement is measured across recommendation outputs, not across hidden reasoning or retrieval context.
Average agreement can hide large differences between query classes and model pairs.
The study measures overlap, not which answer was objectively “right”.
What to do with this
- Track visibility across multiple models instead of using one platform as a proxy for the whole market.
- Prioritize query classes where disagreement is highest because that is where share can move fastest.
- Treat consensus as a benchmark but treat divergence as the operating reality.
Relying on a single model for visibility metrics leads to blind spots in resource allocation. Teams must use multi-model tracking to accurately decide which content to publish, refresh, or measure next.
Questions about this answer.
Cite this answer.
Source study: Model Divergence“No. With only 43.3% average agreement and 4.0% perfect consensus across 8 models, one model is an unreliable proxy for the wider AI market.”
Mack Grenfell · Should you use one model as a proxy for all AI visibility?
Same Question, Different AI, Different Answers · Trakkr Research · Updated 11 March 2026
Apply the finding
Model divergence changes both answer targeting and generative optimisation. Separate the AEO and GEO work this finding affects before assigning the next action.
Because one model cannot proxy for the rest, measure Google AI Overviews as their own surface instead of inferring Google visibility from another assistant.