AI buying agents ignore scarcity timers. Star ratings are the only cue that moves them.
A Bayes/King's College study of simulated shopping trips found classic persuasion tactics flat or counterproductive on four leading models. Gartner expects agents to intermediate 90% of B2B purchases by 2028.
Across thousands of simulated shopping trips on four leading AI models, researchers at Bayes Business School and King’s College London found that star ratings were the only cue that consistently moved an agent’s choice. Scarcity messages and countdown timers did little reliably, and sometimes backfired. Jafar Sabbah and Oguz A. Acar published the findings in Harvard Business Review in May 2026 under a title that doesn’t bury the lede: “Traditional Marketing Doesn’t Work on AI Shopping Agents.”
The timing isn’t incidental. Sabbah and Acar point to OpenAI’s deeper push into product discovery, Google’s universal commerce protocol, and Amazon’s tools that let its agents shop other retailers’ sites. Agents are, in their words, “researching, comparing, and increasingly purchasing on behalf of consumers.” Gartner, presenting at its October 2025 IT Symposium, forecasts that 90% of B2B buying will be AI-agent intermediated by 2028, with Digital Commerce 360 reporting more than $15 trillion in B2B spend routed through agent exchanges on that timeline.
Writing in Forbes on October 20, columnist Stephen Wunker frames the squeeze as a split between the cheapest offer and the best fit, with the middle eroding. His operative test for any small business: “Can an agent verify our differentiation from data it can look up?”
That question reorganizes a decade of small-business marketing orthodoxy. Scarcity copy, urgency banners, and brand-warmth tactics were tuned to human loss aversion. An agent reading a product page as structured data doesn’t flinch at a red timer. It parses the rating, the spec, the price, and moves on.
This is the third signal in a week pointing in one direction. B2B buyers now start with an AI agent rather than a search engine. Seven in ten AI shoppers plan to use AI for holiday discovery. The agent layer keeps getting earlier in the funnel.
The persuasion stack built for human attention is being quietly delisted by the software that now stands between a seller and a buyer.
Sources
- https://www.forbes.com/sites/stephenwunker/2026/10/20/ai-agents-will-split-markets-between-the-cheapest-and-the-best-fit/
- https://hbr.org/2026/05/research-traditional-marketing-doesnt-work-on-ai-shopping-agents
- https://www.gartner.com/en/newsroom/press-releases/2025-10-21-gartner-unveils-top-predictions-for-it-organizations-and-users-in-2026-and-beyond
- https://www.cnbc.com/investingclub/2026/10/09/ai-agents-could-soon-shop-for-consumers-what-it-means-for-retail-stocks.html
- https://www.digitalcommerce360.com/2025/11/28/gartner-ai-agents-15-trillion-in-b2b-purchases-by-2028/