How an engine can cite many sources lightly—or rely deeply on a much smaller evidence set.
Breadth and depth can move in opposite directions
In Zhang, He, and Yao’s dataset, Perplexity averaged 16.35 citations per prompt, Google 12.06, and ChatGPT 6.88. Yet ChatGPT showed a substantially higher mean per-source influence proxy among successfully fetched pages.
These figures describe one empirical snapshot, but they expose a durable measurement problem: a longer source list does not establish deeper use of each source.
Breadth and depth answer different business questions
Citation breadth asks how often and how widely a source appears. Citation depth asks how strongly a particular source shapes the answer. Link exposure, factual authority, brand representation, traffic, and conversion can therefore point in different directions.
The selection–absorption paradox
Consider one platform showing twenty lightly used citations and another integrating six sources deeply. The first may create more link exposure; the second may create more answer influence.
There is no single best platform until the visibility objective is defined.
Use a visibility vector instead of one score
Martinez separates discoverability, context exposure, citation, prominence, absorption, fidelity, and behavior. Reporting these dimensions separately prevents a high citation count from masking weak answer participation—or strong participation from being mistaken for traffic.
Questions about this topic
What is citation breadth?+
It is the number or diversity of visible sources cited across an answer or prompt set.
What is citation depth?+
It is the degree to which an individual source contributes to the answer’s facts, language, evidence, or structure.
Does more breadth mean more influence?+
No. An engine may distribute the answer across many lightly used sources while another relies more heavily on a few.
Which metric matters more?+
It depends on the objective: source exposure, answer influence, referral traffic, brand representation, or conversion.
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
- 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