A SIGIR Study Measures Which of Two Sources Is Cited First

What Gets Cited: Competitive GEO in AI Answer Engines is listed as a SIGIR 2026 paper by Rahul Vishwakarma, Shushant Kumar, and Ratnesh Jamidar, affiliated with Sprinklr. The proceedings line places it at the 49th International ACM SIGIR Conference, 20-24 July 2026, in Melbourne, and gives the DOI 10.1145/3805712.3808445. The abstract says AI answer engines generate answers from retrieved pages but cite only a few sources, so visibility depends on being cited, not only on rank.

The testbed injects exactly two candidate sources into the model context and measures which source is referenced by the first citation marker. Across six LLMs the authors report 252,000 trials, repeated paired comparisons under one factorial program over 18 content factors. In each trial the two sources differ in exactly one factor. Brand anonymization and counterbalanced source order are used to separate content effects from position bias.

Mixed-effects models in the abstract show that topical relevance and list position are the biggest drivers of being cited first. Including explicit price information and a recent timestamp also helps consistently. Completeness and trust cues add smaller gains. Formatting-only edits have little impact. The abstract says the authors release a reproducible evaluation protocol and a prioritized GEO checklist, and that an early internal pilot at Sprinklr produced positive qualitative feedback on workflow usability. That pilot sentence is about usability feedback, not a measured traffic lift.

What the abstract does not claim

The opening pages frame citation selection as a visibility bottleneck inside retrieval-augmented generation. They do not report an organic ranking change on a public search engine, and they do not assign a percentage gain to any single rewrite. The unit of the experiment is the first citation marker when two already retrieved documents compete.

Source: Vishwakarma, Kumar, and Jamidar, SIGIR 2026.