---
title: "What a Merchant Protocol Governs"
description: "A merchant protocol governs the product facts a company controls. A shopping agent recommends from the open-web sources it retrieves."
canonical: "https://jaxonparrott.com/blog/agentic-merchant-protocol-product-catalog-control-ai-agents-2026"
date: "2026-03-17"
last-updated: "2026-11-14"
---

# What a Merchant Protocol Governs

A merchant protocol governs the product facts a company controls. A shopping agent recommends from the open-web sources it retrieves. Buy the governance when the catalog is inconsistent, and earn the citation when you want the recommendation.

The comparison I keep beside this is [L’Oréal and Unilever Fixed the Wrong AI Visibility Layer](https://authoritytech.io/curated/loreal-unilever-amp-product-data-wrong-layer-ai-visibility).

## What the March 12 launch changed

On March 12, 2026, Azoma launched the Agentic Merchant Protocol as merchant-side infrastructure for how a catalog is defined, distributed, and governed across AI commerce surfaces ([Azoma](https://www.azoma.ai/agentic-merchant-protocol); [Business Wire release](https://finance.yahoo.com/news/azoma-launches-merchant-side-standard-110000769.html)). Azoma says AMP sits above OpenAI's ACP and Google's UCP rather than replacing them ([Business Quarter](https://businessquarter.co.uk/azoma-launches-protocol-for-agentic-commerce/)). The early adopters named with the launch are L'Oréal, Unilever, Mars, Beiersdorf, and Reckitt ([VentureBeat](https://venturebeat.com/infrastructure/how-to-make-your-e-commerce-product-visible-to-ai-agents-use-this-new-system)).

That list is urgency inside large CPG companies. It is a measurement of adoption, and it is not a measurement of what a shopping agent cites. Max Sinclair, Azoma's CEO, put the operating change in the launch release and a VentureBeat interview: marketplaces had controlled product detail pages, rankings, and distribution, and in an agentic setting those fixed pages are no longer the place the recommendation is assembled. The same interview records his point that companies which spent decades on brand equity are unwilling to hand representation of their products to an ungoverned model.

Read that as a governance claim. The company wants one place for catalog data, brand guidelines, compliance rules, regulatory requirements, competitive positioning, and persona-level targeting, then distribution of that package. Azoma's release lists four AMP components: canonical machine-native product catalogues, programmatic open-web distribution, contextual prioritization of content, and an agent-agnostic interface. VentureBeat adds RegGuard, a compliance check, and citation tracking from the wider platform. Those are merchant-side controls. Agent-side protocols (ACP, UCP, AP2, MCP) describe how an agent connects, pays, and reads context. They do not decide whether the fact the agent repeats is the fact the brand would stand behind.

## The market numbers describe shopping, not a SKU

McKinsey's agentic commerce work projects that by 2030 as much as $1 trillion in U.S. B2C retail revenue could be orchestrated through agentic commerce, and that the global opportunity reaches $3 trillion to $5 trillion. [Morgan Stanley Research](https://www.morganstanley.com/insights/articles/agentic-commerce-market-impact-outlook) estimates that agentic shoppers could reach $190 billion to $385 billion in U.S. e-commerce spending by 2030, a likely 10% share of online retail and up to 20%. A [Riskified survey](https://www.riskified.com/press/global-study-73-of-shoppers-using-ai-in-shopping-journey-but-merchants-face-new-agentic-commerce-risks/) of more than 5,000 consumers, published October 2025 by a fraud-prevention vendor, found 73% using AI somewhere in their shopping journey: 45% for product ideas, 37% for summarizing reviews, 32% for price comparison. Only 13% reported completing a purchase after an AI referral, while 70% were at least somewhat comfortable with an agent buying on their behalf.

Those figures justify attention. They name no protocol, no domain, and no SKU, and they do not say L'Oréal's adoption moved a citation rate. The purchase split is the operating fact. Most of the reported AI shopping behavior is still research, comparison, and review summary. Comfort with an agent buying and completion of a purchase sit far apart. The agent is assembling a description, often from review and comparison pages, and only sometimes closing. A protocol that governs the closing payload and never reaches the review page is aimed at the smaller behavior. Fund the description the 45% and the 37% are reading. Treat checkout governance as a separate operations project.

Azoma also reports two client outcomes. VentureBeat repeats them from the company. Neither the release nor other launch coverage I found audits them. Ruroc's founder says the brand is consistently the #1 most recommended ski and snowboard helmet brand in ChatGPT across multiple geographies, and that ChatGPT-sourced site traffic rose 14x. Perfect Ted, a matcha brand, says Azoma's AI-search visibility work contributed to +532% year-over-year revenue growth across all channels. Both are vendor-reported case results. A 14x traffic figure needs a baseline. A 532% revenue figure needs the period, the channel, and what else changed in the business. The launch reporting does not supply that audit. I ask for it in a contract before either number justifies a category bet, and I do not import either percentage into a forecast for a different category.

Google's Universal Commerce Protocol announcement on January 11, 2026, and the Commercetools note on agentic-commerce trends from January 8, explain why every commerce team heard this pitch in the same quarter. Context is not a measurement of a SKU.

## The retrieval pool is still wide

Product visibility and brand visibility fail in the same place when the machine is the reader. The reader changed. The pool of sources did not collapse into the brand's own product page. Measured on the Machine Relations Index, release `mri_score_v2.1+2026-11-04+94f24cb38da0`, the [Consumer Products category](https://machinerelations.ai/index/categories/consumer-products) (1,825 observed runs across six question shapes) has no dominant source. YouTube, the most-cited, appears in 13.7% of runs (250 of 1,825), Reddit in 13.5%, and Forbes in 10.6%. Each question shape's leaderboard holds between 355 and 681 distinct cited domains.

That spread is why "we implemented AMP, so the agents have us" fails. A canonical record that never leaves the vendor's pipe is a governed document. A recommendation is a retrieval event. If no source appears in more than about 14% of runs, one integration starts distribution, and the facts still have to land on pages an agent will fetch. The stack for that work is [Machine Relations](/machine-relations), earned authority through measurement. If I cannot name, after thirty days, which of those domains received our facts, the protocol has governed a record and has not entered the retrieval pool.

## Two budgets

The catalog budget pays for one merchant-side record: attributes an agent can extract, regulatory constraints, and a rule for what may not be said. For a regulated product, that record is the point of a protocol. Inaccurate agent-generated claims are a legal exposure in categories such as supplements. VentureBeat describes RegGuard as auditing content against rules like FDA/DSHEA standards. I buy that governance when the company cannot keep one product fact consistent across the surfaces it already publishes.

The visibility budget pays for the sources in the retrieval pool. Take the buyer questions that already lead to a recommendation, record which domains the answer cites, and put structured, checkable product facts on the domains that recur. A syndication promise that cannot name those domains after a month does not count. OpenAI's ACP and Google's UCP, in the same sequence as Google's Universal Commerce Protocol post on January 11, 2026, are connections. Connections move a payload. They do not choose the sentence the shopper hears.

This applies to companies whose buyers ask an agent what to buy, and whose product facts are split across a site, a marketplace feed, and uncontrolled affiliate pages. The product page remains the record a human can open. On the evidence in this launch, the agent often does not open it. Until a published measurement, with a window and a denominator, shows recommendations in the category resolving from the merchant protocol's record more often than from the open-web domains in that Index release, the protocol stays in operations and the citation work stays in the source layer.

## FAQ

### Does a merchant protocol replace product pages?

No. ACP, UCP, AP2, and MCP connect agents to data, payments, and context. AMP, as Azoma describes it, is a merchant-side layer above those protocols. The product page is still the human-readable record. The protocol is a claim about governing the facts an agent may repeat, and about pushing those facts to third-party sources.

### Do the L'Oréal and Unilever adoptions prove agents recommend from the protocol?

They prove that five large CPG companies, L'Oréal, Unilever, Mars, Beiersdorf, and Reckitt, adopted the protocol quickly. The 14x ChatGPT traffic figure and the 532% revenue figure are Azoma's own client reports. Trade coverage of the March 12, 2026 launch reports the adopter list and the design, not an independent audit of those outcomes.

### Should a smaller catalog adopt the same protocol?

Adopt the governance piece if inconsistent product facts are already causing bad answers, and if the company cannot staff a separate integration for every agent surface. Do not adopt it as a substitute for the sources agents retrieve. The Index release above puts the most-cited Consumer Products source in 13.7% of runs, with 355 to 681 distinct domains per question shape. One feed does not cover that spread by existing.

### What should a vendor show before a signature?

Which domains received the catalog in the last thirty days, which buyer questions cite those domains, which of the reported lifts were measured by someone other than the vendor, and what the compliance check blocks. A protocol that distributes a claim the legal team would not sign is a faster way to be wrong.

---

*Published March 17, 2026*
By [Jaxon Parrott](https://jaxonparrott.com) — Founder & CEO, [AuthorityTech](https://authoritytech.io)
Updated 2026-11-14

Part of [Machine Relations](https://jaxonparrott.com/machine-relations) · [Glossary](https://machinerelations.ai/glossary) · [Founder judgment](https://jaxonparrott.com/topics/founder-judgment)


Tags: Machine Relations, AI Search, Agentic Commerce, Founder Judgment

## Continue reading

- [The Category Ownership Trap](https://jaxonparrott.com/blog/category-ownership-trap): Owning a category name leaves the buyer's question open. I coined Machine Relations in 2024, and the commercial question is still a separate job.
- [Can an AI Engine Name Your Brand?](https://jaxonparrott.com/blog/why-i-check-whether-an-engine-can-name-the-brand): I check whether an engine can name the brand before I buy citations, because a weak identity makes the citation contest a false read.
- [AI Referral Traffic Jumped 527 Percent](https://jaxonparrott.com/blog/the-527-percent-inflection-point): AI-referred sessions on 19 GA4 properties rose 527 percent year over year. I keep SEO and add a citation scoreboard while most teams report the old channel.

---

*Machine-readable version of [https://jaxonparrott.com/blog/agentic-merchant-protocol-product-catalog-control-ai-agents-2026](https://jaxonparrott.com/blog/agentic-merchant-protocol-product-catalog-control-ai-agents-2026)*
