Meta's Shopping Agent Told to Rank Known Merchants First

Meta’s Muse shopping agent must fill its top five product picks with well-known merchants, according to an instruction file Profound says it pulled on September 25.

If you run a high-ticket Shopify store, this is a rare look at how an AI agent decides which stores a buyer sees. For the store owners we cover at Ecommerce Paradise, the catch is that Shopify’s default settings may already be feeding your products to Meta, so the rule applies to you whether you opted in or not.

Below: what the file says, how Muse got here in five weeks, what the known-merchant rule does to a MAP-priced store with a thin brand footprint, and five changes to make before the holiday rush. I covered an earlier study of AI shopping picks favoring big retailers, and this file offers a possible explanation of why.

Agents sort stores by trust, and a properly formed LLC is the first trust signal you control, so form yours with Northwest Registered Agent, the best LLC formation service for ecommerce owners. Northwest has run for 25 years under stable ownership, which is why it beats the cheap formation sites that upsell you every year. Form your LLC with Northwest →

Meta Muse Shopping File Ranks Well-Known Merchants First

PPC Land reporter Luís Rijo published the findings on October 11, building on a State of Brand analysis from October 8. Profound, a company that sells software tracking how AI agents treat brands, says it pulled Muse’s shopping instruction file from the agent on September 25 and shared it with State of Brand. Profound warns the file may have changed since then. Meta has not commented.

Per the report, the file calls it “imperative” that Muse’s top five products come from well-known merchants or websites. It does not define “well-known.” PPC Land notes it is unclear whether Muse relies on Meta’s own data, web signals, or the model’s general knowledge to decide.

Four other rules matter to a store owner. Listings priced far above or below a category’s normal range are removed, and the report says low prices, often used by newer brands, get filtered too. For each request Muse searches the open web on up to three merchant sites and returns up to ten products, and it may stop early if two sites supply enough. When a shopper names a brand, that brand’s own store ranks first, and direct-to-consumer brands are favored by default. In-chat checkout and carts work only for Meta catalog products carrying an “agentic checkout” flag, while open-web items use browser checkout and show product cards with in-chat checkout disabled.

Profound paired the file with a test. In a second PPC Land report on the research, Profound’s team described running 50 prompts in one consumer retail category through a single Muse instance on September 30. The medians: six final recommendations, four from browsed sites and two from Meta’s catalog. A typical catalog search returned about 60 products from 30 merchants, while the browser visited three domains and found four products at two merchants. Profound calls the sample “a small, exploratory sample” whose patterns are “illustrative rather than definitive.”

The audience is not small. Sensor Tower estimates more than 5 million US and Canada installs within 22 days of launch, per the PPC Land report. CBS News reported over 2.5 million downloads in about two weeks, also citing Sensor Tower, against 3.1 million global downloads for ChatGPT over the same stretch. Meta has released no figures of its own.

One more constraint from the checkout side. Steve Hutt, a former Shopify Merchant Success Manager, wrote in an October 6 eCommerce Fastlane guide that per Shopify’s documentation, subscriptions, bundles, customizable products and B2B-only products cannot be bought through Meta direct checkout, and that Google Analytics and custom pixels do not fire there.

From Muse Launch to Amazon Block: How Meta Got Here

Meta launched Muse on September 8, 2026, as a personal agent that can operate a browser, fill out forms and negotiate for users. According to Search Engine Journal’s October 2 breakdown, Meta received eligible Shopify stores’ products that same day. Shopify’s setting “Allow Shopify to manage for me,” under Sales channels and then Agentic, is on by default, and Shopify’s changelog says products are “shared with Meta by default.” Direct checkout for Meta covers stores selling to the US, Canada or Mexico.

I wrote about the Shop Pay side of that deal in how Muse can now check out with Shop Pay. The protocol underneath is in my post on Shopify and Stripe joining Meta’s personal agent protocol.

On September 20, Amazon cut Muse off. An Amazon spokesperson said third-party shopping apps “should operate openly and respect service provider decisions about whether or not to participate,” and that continued access by “an unauthorized AI agent violates Amazon’s Conditions of Use,” per PPC Land’s October 8 report on the block. Amazon said the agent does not identify itself while browsing and appears to capture customer credentials. Meta has said Muse “has no visibility into people’s passwords or payment methods.”

On September 29, Meta announced Muse for Small Business. The Next Web reported it connects the agent to tools owners already use, including Shopify, Stripe, Canva and Intuit’s QuickBooks. Vishal Shah, Meta’s vice president of AI products, told Axios, “People are using Muse to run their business.” Meta says nothing publishes, sends or spends without the owner’s approval.

Shopify’s side of the money is quiet for now. SEJ reports that as of September 2026, ChatGPT, Google, Copilot and Meta charge merchants no fees beyond normal payment processing, and Shopify merchants stay the merchant of record. Mark Zuckerberg said on the Sources podcast that Meta expects to take a very small cut of Muse transactions, coming from the businesses involved.

The evidence has real limits, and they cut both ways. Profound sells AI visibility products, so it has a commercial stake in findings that say brands need help being seen. The test was one account, one category and one day, and nobody has replicated it. PPC Land also notes that Sensor Tower reported Meta ramping Muse ad spend on September 16, which could explain part of the traffic climb Profound measured. And the report observes that a human merchandiser would make many of the same choices. The difference is that Muse applies them automatically to every request.

What Muse’s Known-Merchant Rule Does to Your Store

My read: “well-known” will behave like a brand-search proxy, whatever Meta calls it internally. If nobody types your store name into Google, an agent that has to rank five picks fast has little reason to put you there. That is a bad fit for a new high-ticket store, because the whole model is a niche site that wins by being the specialist, not the household name.

The price filter is less scary than it sounds, at least on single products. If you sell authorized, MAP-priced inventory, your price sits inside the normal band by construction, which is one more reason I push dealers toward authorized-dealer, MAP-protected high-ticket dropshipping. The risk is the stuff you build yourself. A bundle with installation, an extended warranty and white-glove delivery can land far above the category band, and the file says listings far above it are removed. That is my inference from the reported wording, not something Profound tested.

The catalog is the part you control. Profound found the catalog supplied about 95% of candidates at the median, so being in Meta’s catalog gets a product into the pool, and it enables in-chat purchase. State of Brand adds that catalog integration does not change the ranking rules. Treat Shopify catalog hygiene as table stakes: clean titles, correct GTINs, accurate availability, and prices that match your site to the dollar.

Run rough math, labeled hypothetical. Say a store does 40 orders a month at an $1,800 average order value, so $72,000 in revenue. If agent-assisted checkout took 5% of those orders, that is two orders and $3,600 a month. Not a reason to rebuild anything. Now say the same agent shapes research for 20% of your shoppers who then buy through Google or your site. That influence will not show up as Muse revenue anywhere, which is the part that should worry you.

Two caveats keep this honest. Muse is five weeks old, and one month is not a trend. And the file may already be out of date, since Profound said so itself.

Here is how I’d sort stores by scenario.

If shoppers already search your store name and you have a few hundred real reviews, you are probably fine in the browser path, and your job is keeping the catalog clean. If you are a new brand under a year old, assume you are in the candidate pool and not in the top five, and weigh Meta direct checkout against its limits before you leave it on. If you sell configurable or custom products, subscriptions or B2B-only items, direct checkout cannot sell them per Shopify’s documentation, so the Agentic channel is mostly a visibility play for you.

Tracking is the quiet cost. Per SEJ, orders completed inside a chat app’s checkout do not fire your site’s tracking code, so they can be missed in Google Analytics or ad dashboards while still appearing in Shopify admin. If your Klaviyo flows and ad reporting key off site events, test a real order before you assume post-purchase email and attribution still work. And if you let Muse read your books through QuickBooks, QuickBooks will show the revenue even when your ad dashboards do not.

The broader pattern is the one I flagged when Shopify started showing AI agents your compare-at prices: agents read the data you already publish and act on it. Messy data was a conversion problem. It is turning into a visibility problem.

The same logic applies to paid channels, where ChatGPT ad clicks measured at six times Google’s cost. Google’s side points the same way, since the UCP checkout hub in Merchant Center also runs on your structured product feed.

This is also where the operator workload piles up. Catalog audits, pricing checks across channels, agent checkout settings and tracking tests are four small jobs that add up to a part-time role. If you would rather hand that to a team, my turnkey done-for-you build sets the store up with clean feeds from day one. Stores that already have revenue can use my scaling service instead.

Agents will shop for your customers whether your catalog is clean or not, so let my team build and run a high-ticket store that holds up when they do. See the turnkey done-for-you service →

Fix Your Muse Settings and Price Signals This Week

Five actions, in the order I would do them:

Open Shopify admin, go to Sales channels, then Agentic, and decide on “Allow Shopify to manage for me.” Turning it off reveals three switches (Shopify Catalog access, Auto enroll new storefronts, and direct checkout), and per SEJ catalog sharing can take up to seven days to stop. Keep the catalog on, and make a deliberate call on Meta direct checkout instead of inheriting the default.
Pull your top 20 products and compare each price to the three closest competitors. Flag any bundle or package that sits far above its category band, and list the underlying single products separately so an agent has a clean, in-band item to pick up.
Build your “known” signals now. Claim and complete your Google Business Profile, collect reviews on a platform buyers recognize, and make sure a search for your exact store name returns your site first. The file does not say these count, so this is my read on what a brand-search proxy would pick up. If you want a second pair of eyes on your brand footprint, book a call through my discovery page.
Test a real order. Buy one low-risk item through a chat checkout if you can, then confirm it appears in Shopify admin, triggers your Klaviyo flow, and shows up in the right place in your books. Write down which dashboards miss it.
Run a monthly agent check. Ask Claude and ChatGPT to recommend stores for your top product, and record whether you appear. Track movement over time with the brand and keyword reports in SEMRush.

Before you connect any Meta tool to your ad account, read my breakdown of Meta’s ban on TikTok ads and decide who inside your business can approve actions. PPC Land notes Meta has not said whether Muse can only read ad performance or can also change budgets, bids, targeting or creative. Until that is published, I would leave ad accounts out of any Meta agent connection and add a second connector only after 30 days of clean results.

Frequently Asked Questions

What does the leaked Muse shopping file say?
Per PPC Land, it calls it “imperative” that Muse’s top five products come from well-known merchants, removes listings priced far outside a category’s normal range, and ranks a named brand’s own store first. Profound says it pulled the file on September 25, and it may have changed.

Has Meta confirmed the file is real?
No. Meta has not commented, per PPC Land. Treat it as unverified evidence from a vendor with a commercial interest, and build for the rule being roughly true.

Is my Shopify store already sending products to Meta?
Probably, if you are eligible and left defaults on. SEJ reports “Allow Shopify to manage for me” is on by default, and Meta received eligible stores’ products on September 8. Check the Agentic settings in your admin, and see my post on Shopify showing AI agents your compare-at prices.

Can a new store still get picked?
It can enter the candidate pool through the catalog, but the reported rule favors known merchants for the top five. Start with the niche work in my free high-ticket niches list and build brand searches before you chase agent traffic.

Does Meta charge merchants for Muse orders?
As of September 2026, SEJ reports no fees beyond normal payment processing. Zuckerberg has said Meta expects to take a very small cut from the businesses involved, so plan for that to change.

Will Muse buy from Amazon?
No. Amazon blocked Muse on September 20, citing its Conditions of Use and credential handling, and said it is talking directly with Meta.

Do chargebacks change with agent orders?
Meta has not published dispute rules for Muse orders. Until it does, keep your processes tight using my guide to stopping disputes on high-ticket orders.

Want to work through your agent settings and catalog with other store owners and me inside the community? Join the Skool community →

I will keep tracking what Profound and State of Brand publish next, and I will update this if Meta responds. If you run a processor behind your Shopify checkout, also read why one payment processor is a single point of failure. Subscribe to the YouTube channel for daily breakdowns. More breaking news coming through the day.

If this was useful, these go deeper:

Trevor Fenner

Trevor Fenner is an ecommerce entrepreneur and the founder of Ecommerce Paradise, a platform focused on helping entrepreneurs build and scale profitable high-ticket ecommerce and dropshipping businesses. With over a decade of hands-on experience, Trevor specializes in high-ticket dropshipping strategy, niche and product selection, supplier recruiting and onboarding, Google & Bing Shopping ads, ecommerce SEO, and systems-driven automation and scaling. Through Ecommerce Paradise, he provides free education via in-depth guides like How to Start High-Ticket Dropshipping, advanced training through the High-Ticket Dropshipping Masterclass, and fully done-for-you turnkey ecommerce services for entrepreneurs who want a faster, more hands-off path to growth. Trevor is known for emphasizing sustainable, real-world ecommerce models over hype-driven tactics, helping store owners build scalable, sellable, and location-independent brands.