Three ways to measure visible sources—and why none of them alone captures source influence.
Citation breadth counts distinct sources
Citation breadth is an answer-level measure: how many sources does one response expose? Zhang, He, and Yao reported mean citations per prompt of 6.88 for ChatGPT, 12.06 for Google, and 16.35 for Perplexity in their dataset.
These are platform snapshots, not permanent laws. A broader source list does not prove deeper use of each source.
Breadth and depth can move in opposite directions
The same study found higher average per-source influence for ChatGPT under its constructed proxy despite fewer citations. The useful conclusion is not that one engine is better; it is that citation count and source influence are different dependent variables.
How many sources are shown and how much each source shapes the answer are different questions.
First citation position measures prominence
A source appearing near the opening may receive more observable exposure than one listed only at the end. Useful measures include first citation position, normalized first-position ratio, repetition, top-third presence, and first-source share.
Position is a prominence measure. It does not directly prove user attention or internal source influence.
Domain frequency reveals recurring gateways
Counting how often domains recur across prompts helps map source ecology. High-frequency domains repeatedly qualify for visible selection, but frequency does not establish that they contribute the most facts or language.
Questions about this topic
What is citation breadth?+
It is the number of distinct visible sources cited by an answer, usually measured per prompt or response.
Is more citation breadth always better?+
No. It may improve coverage, but does not establish deeper use, higher accuracy, or greater value from each source.
What does first citation position measure?+
It measures where a source first becomes visible in the answer and is best interpreted as observable prominence.
What does domain citation frequency reveal?+
It identifies recurring source gateways across prompts, not necessarily the sources with the greatest answer influence.
References
Sources are listed in APA 7 style. Preprints are identified as such and should not be treated as peer-reviewed findings unless separately published.
- Zhang, K., He, X., & Yao, J. (2026). From citation selection to citation absorption: A measurement framework for generative engine optimization across AI search platforms [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2604.25707
- Aggarwal, P., Murahari, V., Rajpurohit, T., Kalyan, A., & Narasimhan, K. (2024). GEO: Generative engine optimization. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (pp. 5–16). Association for Computing Machinery. https://doi.org/10.1145/3637528.3671900
- Martinez, O. (2026). Optimizing visibility in generative engines: A critical survey of generative engine optimization (2023–2026) [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2607.14035