Commerce

Profound says Meta’s Muse shopping agent is told to rank well-known merchants first; Meta has not commented

Developing

The instructions are Profound's extraction, not published by Meta, and may have changed since late September.

A woman looking at her mobile phone in a store showroom
Photo: Woman with Phone by epSos.de, CC BY 2.0, via Wikimedia Commons

In 60 seconds

  • Instructions Profound says it extracted call ranking well-known merchants first in the top five picks "imperative".
  • The agent may search three web stores, return up to ten products and offer in-chat checkout only for Meta catalog items flagged for agentic checkout.
  • In a 50-prompt test, a typical query pulled about 60 catalog products from 30 merchants, while the browser found four products at two merchants.
  • Catalog supplied roughly 95% of candidates but won about a third of final slots.

Impact Layer

What changed
There is evidence of how one AI shopping agent weighs brand familiarity against catalog data.
Why it matters
If agents favour known brands and web results, smaller merchants may need web presence as well as feed data.
Who is affected
Merchants, especially smaller ones, and teams working on agentic commerce.
Martech stack area
AI, Commerce
Stack impact
Make sure both your catalog feed and your public web pages are complete, because the agent may weight web search more than catalog data.
Business impact
Moderate Single-vendor finding relayed by PPC Land. Profound calls the sample "small", "well-known" has no stated definition, and Meta has not commented.
What to watch
Meta's response and independent tests of agent ranking.

PPC Land reported on October 12, 2026 on analysis by Profound of Meta’s Muse shopping agent.

The extracted instructions

  • Instructions Profound says it extracted, dated September 25, call ranking well-known merchants first in the top five picks “imperative”.
  • The agent may search three web stores, return up to ten products, and offer in-chat checkout only for Meta catalog items flagged for agentic checkout.

The test

  • In a 50-prompt test, a typical query pulled about 60 catalog products from 30 merchants, while the browser visited three sites and found four products at two merchants.
  • The catalog supplied roughly 95% of candidates but won about a third of final slots.
  • Profound reported Meta-WebIndexer requests to 115 US and Canadian sites rose 1,313% over two 20-day windows, against 827% for 165 other sites. It calls the sample “small”.

Profound economist Jennifer Zou: “Much more weight is placed on candidates returned by the web search.”

Business expansion

On September 29, Muse reached small US and Canadian businesses through 15 connectors, including Shopify, Stripe and QuickBooks, plus Meta ad accounts. Meta says nothing publishes or spends without owner approval.

Not stated

How “well-known” is defined, how the three stores are chosen, whether ad account access is read-only, whether the instructions have changed and Meta’s response.

Our read

Treat it as one vendor’s reconstruction. The practical lesson is to be findable in both feeds and on the open web.

Sources

  1. PPC Land: Meta's Muse agent puts well-known merchants first as it gains ad accounts (opens in a new tab) News report ·

    Source for the extracted instructions, test results, quotes and gaps.

Get MarTech Roundup in your inbox

What changed in marketing technology, why it matters, and what to do next.

We will email you to confirm. You can unsubscribe at any time.

Related stories