Study prices sellers' hidden influence on AI shopping agents
What this answers
- Can sellers manipulate what an AI shopping agent recommends?
- Yes — a new study finds seller-controlled cues like badges and endorsements measurably shift which product an AI shopping agent picks, with the effect large enough to express as a dollar premium.
- Are all AI shopping models equally swayed by seller cues?
- No. Jev, a decision-native model tested in the study, resisted most seller cues but still gave in to the strongest social-proof claims — a different profile than any text-based model tested.
A study of 113,472 trials across seven AI models finds that seller-controlled content — badges, endorsements, framing — shifts AI shopping agents’ product picks by amounts that vary widely even when two cues look equally persuasive by simple win-rate. The authors introduce “price-equivalent auditing,” which raises a cued product’s price until the model picks it only half the time, to measure that gap in dollar terms. Tested against a decision-native model, Jev, which outputs a choice probability rather than text, the authors found it resisted most cues but remained susceptible to the strongest social-proof claims — a profile unlike any text model in the study.
Why it matters: Choice-rate audits saturate once a cue wins almost every time, hiding how much stronger one seller tactic is than another; pricing the cue in dollars gives GEO auditors a way to actually rank seller tactics by strength.
Via arXiv GEO ↗
Posted to the wire October 19, 2026. Edited by Joe Balewski.