Is your domain in the retrieval path? That's the difference between 2.8% and 91.4% mention rate
When neither the brand nor its own domain appears in the retrieval path, mention rates sit at 2.8% (GPT) and 3.8% (Gemini). With an own-domain citation, the same prompt reaches 49–58%.
Key numbers
- Sample: 34,960 unbranded prompt-engine observations from 75 projects, covering 2,854 distinct prompts, with repeated GPT and Gemini runs from June to September 2026.
- No brand and no own-domain exposure: 2.8% mention rate on GPT, 3.8% on Gemini. An own-domain citation without branded fan-out lifts that to 49.0% and 58.4%; both together reach 91.4% and 100%.
- Within the same project, prompt and engine across repeated runs: a previous non-mention plus no current own-domain exposure yields next-run mention rates of 1.6% / 1.9%, while a previous mention plus current exposure yields 80.5% / 83.7%.
- Modelling prior-run history together with contemporaneous retrieval signals reaches AUC 0.963 (GPT) and 0.942 (Gemini) on a 30% holdout — prior results alone are predictive.
Method and sample
arXiv paper “From Prompt to Recommendation: A Fitted Stage Model of Brand Visibility in AI Search” (2026-09-19): 34,960 unbranded prompt-engine observations across 75 projects and 2,854 prompts, with repeated GPT and Gemini runs, relating retrieval-path signals (own-domain citation, branded fan-out) to mention rates. The authors state the model is predictive and observational, not causal.
What it means for brands
Whether AI mentions you starts with whether retrieval can reach you: if your site and brand facts are not in the sources AI can retrieve, nothing else matters. Visibility also carries inertia — a run with no mention and no exposure is very likely to repeat.
How SourceSight handles it
Every SourceSight project checks whether the site is cited and which sources appear in answers, and records mention rate per prompt per snapshot. Prompts that stay unmentioned across runs with no own-domain citation are the ones to fix first.
Limits
The panel is English-language and skews to one monitoring dataset; the relationships are correlational, and engine retrieval and ranking internals are not public, so the exposure-to-mention mechanism cannot be verified.
References
Run it on your own project
What the same prompts return across 10 engines is only knowable by running them.
Related research
What 2 million AI citations show: prompt-content match matters most, checklists matter less
Once you control for the domain, the usual AEO checklist (FAQ blocks, structured data, Core Web Vitals) reverses or collapses to zero — domain authority and prompt alignment dominate.
How engines work · 2026-10-07Whose side is the AI shopping assistant on? Change its role and the penalty for sponsored results disappears
The same sponsored listings get treated very differently once the prompt names the platform as the agent's principal — disclosure rules built for humans do not carry over.