Split-Perception
Your brand is visible to AI. Is it being recommended?
Citation is not recommendation. VeritasLinks measures the gap between models to reveal where your company stands in the AI decision room.
Step 1 — Enter your website
Free · Full report on your company
Risk Signal
Visibility is not recommendation.
You see your brand in AI answers, but that presence does not guarantee preference. Citation is not recommendation. Companies can be highly visible yet barely preferred by the models that shape real decisions 1.
Citation is not recommendation.
This gap matters most for risk and compliance teams. Commercial real estate due diligence groups are already using AI for lease abstraction and document review 2. They need to know if the AI judgment is stable or just a random draw.
Models disagree. The spread between different models is a risk signal. A company that is not on the short list does not rank low. It is simply not in the room when decisions are made.
Verifiable preference separates signal from noise. You need to know which score to trust for due diligence. The models project a perception that shapes purchases and partnerships.
Model Disagreement
A 72-point gap between models signals risk in AI due diligence
72
Point Spread
The difference between the highest and lowest model scores for the same company reveals how much perception varies.
741
High Score
DeepSeek and Perplexity rate the company as strongly recommended.
669
Low Score
Grok rates the company lower, indicating less perceived credibility.
15.05
Instability Index
Grok shows the highest run-to-run variation, suggesting less stable answers.
Check if your company holds a consistent recommendation across models or if it is just occasionally mentioned.
Risk Signal
Visibility is not recommendation. Preference is the risk signal.
M&A and risk teams rely on AI outputs for due diligence. But a model can mention a vendor without preferring it.
01
Models disagree on the same company
Different models judge the same company differently. This gap is a measurable risk signal 3.
02
Single pass reports random noise
One model output is just one random draw. It is often reported as a stable fact when it is not.
03
Missing queries hide revenue
Models skip companies in real buyer questions. This leaves revenue at risk without the company knowing why.
04
EU compliance demands monitoring
The EU AI Office has taken action against non-compliance since August. Self-reported data is no longer enough 4.
05
Invisible is not low ranking
A company not on the model's short list does not rank low. It is simply invisible to the decision.
Free analysis
Verify your AI perception today
Get a free multi-model analysis to see if your company is in the room.
Enter your website
Free · No credit card
Perception vs Visibility
Your brand is visible to AI. Is it being recommended?
Organizations are delegating due diligence and vendor selection to AI agents, creating a risk of 'perception drift' where models disagree on credibility despite stable company performance 5.
Get your first analysis free to see if your brand is being chosen or just mentioned.
Risk Signal
Visibility does not mean preference
M&A analysts and KYB reviewers see brands in AI answers but struggle to verify if those mentions drive actual decisions. The models disagree, and a single visibility score often masks this instability. This gap creates risk when stochastic model outputs shape high-stakes judgments without clear stability.
The data reveals a stark disconnect between how strongly models surface a brand and how often buyers choose it. AI-visibility vendors analyzed in July 2026 show high visibility scores alongside low preference metrics. Semrush leads in visibility with a score of 135, yet its buyer preference sits at just 3. Profound follows with 130 in visibility but only a 5 in preference.
This Split-Perception metric measures the gap between different models' judgments of the same company as a risk signal 7. A company can be highly visible and barely preferred. VeritasLinks separates this signal from noise by running many prompts across several models. The result is a clearer view of where a company actually stands in the minds of AI-driven buyers and reviewers.
Visible to the models, rarely picked by buyers
AI-visibility vendors analyzed by VeritasLinks in July 2026. Brand visibility is how strongly models surface the brand. Buyer preference comes from AI-simulated focus groups choosing a vendor.
Source: VeritasLinks analyses, July 2026
Split-Perception
The 72 Point Gap Between Mention and Judgment
What the Reader Sees
“"DeepSeek and Perplexity both know us. We are safe."”
What the Model Judges
DeepSeek scores the company at 741. Perplexity scores it at 741. This agreement looks stable. 8
False ConsensusWhat the Reader Sees
“"Claude and Grok agree. Our brand is solid."”
What the Model Judges
Claude scores the company at 694. Grok scores it at 669. The 25 point gap is often ignored. 8
Hidden SpreadWhat the Reader Sees
“"Our AI visibility is 63%. We are recommended."”
What the Model Judges
A single high score from one model is one random draw. It does not guarantee the company is preferred by the broader model ecosystem. 9
Snapshot BiasWhat the Reader Sees
“"We are visible in search. The models see us."”
What the Model Judges
VeritasLinks analyzes the company across ChatGPT, Claude, Grok, Perplexity and DeepSeek. It shows the separate score for each model.
Visibility GapCheck the spread between models before you trust the score.
Metrics defining the AI judgment risk signal
- VeritasLinks measures Split-Perception to quantify the gap between models judging the same company as a risk signal.
- The Spread metric tracks how much per-model scores diverge during multi-model analysis.
- Missing queries identify real buyer questions where models did not pick your company.
- Estimated revenue at risk is calculated by splitting missed queries by search intent.
- AI Investor Review shows how a credit officer or M&A analyst would read your company's stability.
Questions
Short clarifications — positioning and proof live in the sections above.
Is a high AI visibility score the same as a recommendation?+
No. Citation is not recommendation. Companies can be highly visible but barely preferred by models. VeritasLinks measures how large language models perceive, describe and judge a specific company. It analyzes a company across several AI models and shows a separate score for each model. This distinction ensures that visibility does not automatically equate to strong model endorsement.
How does model disagreement affect due diligence?+
Model disagreement signals risk. VeritasLinks uses a Split-Perception metric where the gap between different models judging the same company is a risk signal. Organizations are delegating due diligence to AI agents, creating a risk of perception drift where models disagree on credibility despite stable company performance. This gap helps analysts identify where AI views diverge significantly.
What is the difference between Split-Perception and standard GEO?+
Split-Perception measures the gap between different models judging the same company. Standard GEO often relies on single model answers or aggregated visibility. VeritasLinks analyzes a company across several AI models and shows a separate score for each model. Each analysis runs many prompts across several models with repeats and measures the spread of answers instead of reporting a single model answer as a fact.
Can I trust a single AI answer for compliance decisions?+
Not necessarily. A single answer may not reflect the full spread of model perception. VeritasLinks analyzes a company across several AI models and shows a separate score for each model. Each analysis runs many prompts across several models with repeats and measures the spread of answers. This approach captures the variability in how different models perceive the same company, providing a more robust basis for decision-making.
Does VeritasLinks measure emotional tone as well as facts?+
Yes. The analysis includes the emotional tone each model carries about the company. This is known as the Perception Affect Field. VeritasLinks measures how large language models perceive, describe and judge a specific company. By capturing this emotional layer alongside factual judgments, the analysis provides a more comprehensive view of how AI models view a company beyond just its structural attributes.
Verify your AI perception today
Get a free multi-model analysis to see if your company is in the room.
- →Same product path as the homepage — enter a public URL to start.
- →Structured interpretation, not a one-off chat screenshot.
Start
Run your analysis
We gather public context, then run comparable model reads.
Step 1 — Enter your website
Free · Full report on your company
Sources
- Google AI Overviews are basically legal content theft at this point · old.reddit.com
- Educational Workshop: AI in Due Diligence · bisnow.com
- Four AI Models, Zero Disagreement: ChatGPT, Gemini, Perplexity and Google AI Mode Agree 100% on Singapore’s Safest Banks · democratandchronicle.com
- The EU is investigating OpenAI and Anthropic following incidents involving AI agents · computerworld.com
- Gemini 4 Argon vs GPT-6 Astra: Real Cost per Task Alternatives · memeburn.com
- Recommended by AI Reveals the Five Rules That Decide Which Brands ChatGPT, Claude and Google AI Overviews Recommend · streetinsider.com
- The new economics of enterprise AI · technologyreview.com
- VeritasLinks report (model_scores) · veritaslinks.com
- I made ChatGPT, Claude, and Gemini summarize the same 200+ page PDF, and the results weren't even close · xda-developers.com