The 5C’s of Agentic Commerce

Agentic commerce optimization category leader Azoma, in partnership with the Digital Shelf Institute (DSI), today announces the launch of “The 5 C’s of Agentic Commerce” report.

This is the first industry standard to pull brands out of fragmented “pilot purgatory” of endless, isolated AEO/GEO experiments and on the path to a unified, profitable commerce strategy.

Recognised by global retail giants Already adopted by Mars, Unilever, Beiersdorf, L’Oréal, and Reckitt, the framework defines five pillars critical to AI-driven retail: Completeness, Context, Citations, Correctness, and Customer Acquisition. The 5Cs have been instrumental in navigating the AI era.

They gave our team a clear, practical framework at a moment when the rules of ecommerce are being rewritten in real time.

– Erica Meagher, Head of Ecommerce, Beiersdorf Canada Agentic commerce is platform-specific The 5C’s of Agentic Commerce framework was developed based on Azoma’s proprietary citation data which analysed tens of millions of AI responses in Q2 of 2026.

The findings shatter the myth of a “one-size-fits-all” AI strategy, revealing how different AI agents lean on vastly different sources to recommend products: Earned media dominates voice agents: Earned media drives 86. 5% of citations for Amazon’s Alexa for Shopping and 76% for Walmart’s Sparky.

LLMs lean on retailers: OpenAI’s ChatGPT and Google’s Gemini rely far more heavily on direct retailer sources (37. 1% and 41. 6% respectively). The category flip: The source mix shifts drastically by aisle. Earned media drives 67. 6% of citations in the Wellness category, while retailer sources dictate 50. 5% of answers in Food.

The hidden scale of the data problem Azoma’s citation data exposes a massive backend hurdle: a single product listing on a global retailer can now require 600+ unique attribute values to be deemed “complete” by an AI.

Because AI agents rely on backend metadata rather than consumer-facing marketing copy to determine product eligibility, brands missing these hidden data points are being locked out of AI recommendations entirely.

Due to AI agents predicting rather than retrieving information, they frequently hallucinate product specs – even when a brand’s official content is 100% accurate. Azoma warns that the only way to catch and correct these fabrications is through systematic, proactive platform testing.