What is Verbatim citation?
A verbatim citation is the synthesis-stage outcome where the engine renders a quoted span from the source chunk directly inside the rendered answer, with a numbered source chip attached to the quoted text. Verbatim citations are the highest-value synthesis disposition because the quoted span is what the user reads as the engine's evidence — the source chip rides with the prose the user is already attending to rather than sitting in a separate 'further sources' panel. The verbatim slot is selective: only 22–34% of chunks surviving rerank earn it in mid-2026. The five properties that lift the verbatim rate are claim specificity, self-containment, rationale-shaped opening sentence, source authority signaled by Article schema and author credentials, and lexical distinctiveness the engine cannot easily compress into its own voice. Verbatim citations also drive 2.3–3.1× the click-through of paraphrased equivalents at the same answer-position.
How it relates to AI UGC
Verbatim citation on multimodal-active sub-queries is the single highest-correlation predictor of inline-carousel inclusion for the page's visual layer — the engine binds the image selection to the cited paragraph at composition time. Pages with verbatim citations win the carousel slot at 2.6× the rate of pages with only paraphrased citations on the same sub-query.
Key statistics
- Verbatim-cited chunks earn click-through at 2.3–3.1× the rate of paraphrased chunks at equivalent answer-position (citation-disposition CTR audits, 2026).
- Verbatim citation rate sits at 22–34% of rerank-surviving chunks on mid-market programs and 38–52% on category-leading programs (verbatim-rate cohort audits, 2026).
- Chunks with explicit numeric statistics (e.g., '38% lift', '8–12 week window') earn verbatim citation at 2.4× the rate of vaguely phrased equivalents because the synthesis stage renders the numeric directly rather than paraphrasing it (claim-specificity-to-verbatim audits, 2026).