Meta Told Its Shopping Agent to Rank Well-Known Merchants First. Challenger Brands Start Behind.

Muse's shopping instructions favor familiar sellers, cut prices that fall outside the norm and check no more than three stores on the open web. The rules make it harder to win the way challenger brands have always won.

For a decade the challenger brand's playbook ran on discovery. Build a better product or price it lower, get found in search or on Instagram, and take share from the names everyone already knew.

Meta's Muse works against nearly every step of that. The agent's shopping instructions tell it three separate times to put well-known merchants at the top of its recommendations. It throws out listings priced outside the normal range for a product, checks no more than three stores on the open web per request and saves in-chat checkout for products in Meta's own catalog. A newer brand can still make the list. It just starts behind.

Profound, an AI search visibility company, says it pulled the file from Muse on September 25 and shared the full text with State of Brand. Profound sells software that tracks this kind of agent behavior for brands, so it has a stake in marketers paying attention. The file is worth reading anyway.

Muse has been downloaded more than 2.5 million times since it launched on September 8, Profound reported, citing CNBC. On September 28 Meta started selling the same agent technology to businesses. Whatever rules sit inside Muse now decide what a lot of shoppers see first.

Muse pulls most options from Meta's catalog and most picks from the web

Profound published research on Muse and Instinct, a personal agent that works over text messages, on October 2. It ran 50 prompts from one retail category through a single Muse account and calls the sample small and exploratory.

On a typical prompt, Meta's catalog returned about 60 products from 30 merchants. Muse's browser visited three sites and found four products at two merchants. The six products Muse ended up showing usually split four from the browser and two from the catalog.

In the median case the catalog supplied about 95% of the candidates, probably because it's cheaper to search, said Jennifer Zou, an economist at Profound with a PhD from Harvard. "Much more weight is placed on candidates returned by the web search," she said.

The instruction file goes a long way toward explaining that gap. Profound said its research team found the file in Muse's system files and that it may have changed since late September.

Brand recognition is a ranking rule

Muse's first instruction is to save the shopper money. When it reviews results, it throws out anything that doesn't match the request, doesn't look high quality or sits outside the normal price range for that kind of product. A listing priced far below the category is as likely to get cut as one priced far above it, and far below is often exactly where a challenger prices to get noticed.

Then it ranks what's left, and the file keeps coming back to the same criterion. Well-known sellers are named in the ranking step of the discovery workflow. The instructions for the results display call it "imperative" that the top five products come from well-known merchants or websites and match what the shopper asked for. The formatting rules say it again.

Being asked for by name pays off as well. If a shopper names a brand or retailer, Muse's catalog search puts that brand's own store at the top and fills in other sellers below. By default it also prefers brands and retailers selling direct over resellers.

A good human merchandiser would make most of the same calls. Muse makes them for every shopper, every time, before anyone has seen an option. That suits incumbents fine. For a brand still building its name, being less known used to cost some clicks. In Muse it costs rank, and the ranking happens before the shopper sees anything.

Meta keeps the register for itself

Products in Meta's catalog that carry an "agentic checkout" flag can be bought inside Muse through a Shopify-based purchase flow. Anything Muse finds on the open web has to go through browser checkout, with the agent clicking its way through the merchant's site, and its product card has in-chat checkout switched off.

Carts work the same way. Muse will build one for a shopper to come back to, but only from flagged catalog products. "Products without that capability have no cart, and neither does browser checkout," the file says.

The catalog is Meta's home turf. It runs shopping on Instagram and Facebook, and the file rates its coverage as good for fashion, home decor and beauty and only okay for everything else. Muse can pull Facebook Marketplace listings too. If a shopper shares an Instagram post with tagged products, Muse goes straight to those items and is told to skip search entirely.

So a brand on the open web can still get recommended. Getting bought inside Muse takes a listing in Meta's catalog.

Outside the catalog, Muse checks three stores and stops

Brands that aren't in the catalog still get a shot. Muse is told to run a browser search next to the catalog search on every request, "especially for home goods," unless the shopper only wanted Marketplace listings.

The search is small. By default the browser agent looks at no more than three merchant sites and brings back no more than ten products, and if two sites give it enough for a "high-quality and seller-diverse" list, it's told to quit early.

Everything it brings back has to pass a few checks. The link has to go to an actual product page rather than a search page, the page has to show the item is in stock and can be added to a cart, and the agent has to be able to verify a real product image. Placeholders and logos get thrown out. Muse also favors the country version of a site that matches the shopper's location, and it reopens every link itself before showing anything.

Nothing in the file says how the agent picks those three stores. A brand that doesn't make that cut never gets compared at all.

Muse knows plenty about the shopper, and nobody has measured what that changes

Before it shops, Muse reads a stored profile of the shopper's preferences, checks a general file about the user and searches its memory for sizes and tastes. For gifts, it goes by what the shopper says about the recipient plus a notes page it keeps on that person.

Some details have to be pinned down before any search runs, like the wearer's gender and size for clothing or the exact device a part needs to fit. Muse asks about them one at a time. The file tells it never to guess someone's gender from their name and never to fill in a default, and a shoe size saved for one brand doesn't carry over to another.

Profound's pitch to State of Brand leaned hard on personalization, but its test used a single account, so it can't say how much two shoppers' results differ for the same request. Zou said the firm is building a panel of synthetic user profiles to find out.

Brands can't see the agent coming

Even when Muse does land on a brand's site, the brand probably won't know it. Profound reports that Muse and Instinct don't identify themselves when they browse. They route through residential internet connections and show up looking like regular visitors, and their answers don't cite the sites they read.

Amazon noticed. Last month it blocked Muse from shopping on Amazon.com, saying in part that the agent didn't identify itself as a bot, State of Brand reported.

Profound tried to estimate the traffic anyway. In its tests, some Muse sessions came right after requests labeled "meta-webindexer" from IP addresses Meta doesn't publicly claim. In Muse's first 20 days, those requests grew 1.6 times as much on U.S. and Canadian retail sites as on sites elsewhere. The comparison covers about 280 sites sorted by domain suffix, and Profound calls it only a proxy.

That leaves marketers with an agent judging their product pages on behalf of real customers and no clear record of it in their analytics.

What a challenger can still control

Zou's advice splits by size. "If you're a well-known merchant/retailer, it's probably fine to rely on discovery via browser search," she said. "If you're a smaller brand, Shopify/Meta catalog integration is a more worthwhile investment."

That's right as far as it goes. For a smaller brand, the catalog is the way into the pool and the only way to be bought or carted inside Muse. It doesn't change the ranking rules, though. Once a product is in the pool, well-known sellers still go to the top.

The rest of the work is unglamorous. Turn on checkout in Meta's catalog. Make product pages easy for an agent to check, with one product per page, a real image and visible stock status. Keep prices inside the normal range for the category, even if undercutting is how you've won before, because the file cuts outliers on both ends. Spell out sizing, fit and compatibility, which Muse has to pin down before it searches. If you sell on Instagram, tag the products, because a shared post with tags skips search altogether.

None of that gets a challenger to the top of the list, and the file is clear about what does. Muse puts well-known sellers first, and it puts a brand's own store first when a shopper asks for it by name. Getting people to ask for you by name is the oldest job in brand marketing. Inside Muse, it's also the most direct way around the rules Meta wrote.