Meta Muse Passes 10M Users; Shopping Is the Top Use

By
Meta Muse Passes 10M Users; Shopping Is the Top Use
Meta Muse Passes 10M Users; Shopping Is the Top Use
Affiliate disclosure: This post contains affiliate links. If you buy through them, I may earn a commission at no extra cost to you. Full disclosure

Meta said Wednesday that more than 10 million people have set up its Muse AI agent, and shopping is the most common thing they do with it.

If you sell online, and especially if you run a high-ticket store like the ones we build at Ecommerce Paradise, this changes who stands between you and your buyer. Meta also told investors it plans to collect a fee on transactions that run through Muse, which means a new middleman is forming in front of your checkout.

I covered the plumbing when Muse gained the ability to check out with Shop Pay. That story had no usage data and no word on a fee. This one has a user count and a stated plan for where the money comes from. Below: what Meta said, what the numbers do and don’t prove, the margin math for a high-ticket store, and five moves for this week.

Agents are coming for your checkout, so make sure the company behind your store is built right. Form your LLC with Northwest Registered Agent, the best LLC formation service for ecommerce owners. If you run your store from outside the formation state or country, Northwest forwards your physical mail, which the cheap formation sites that upsell you never bother to solve. Form your LLC with Northwest →

Muse Passes 10M Users, and Shopping Is the Top Use

The numbers came from Meta’s third-quarter earnings call on October 21, 2026. CEO Mark Zuckerberg said, per the call transcript on The Motley Fool, “In just a month and a half, more than 10 million people in the US and Canada have set up Muse personal agents.” About 25% of people who set up Muse use it on any given day, he said.

On shopping, Zuckerberg was direct: “The most common use case for Muse is shopping.” He put engagement at more than 25% of people using Muse, with “an average purchase size of over $200.” CFO Susan Li framed it differently, saying shopping is “being used by over 25% of Muse weekly actives.”

Those two statements do not describe the same thing. One is a share of people using Muse, the other a share of weekly actives. TechCrunch’s Sarah Perez reported that the $200 figure applies to more than a quarter of engaged users who shopped through the agent, and noted that Sensor Tower had estimated the user count earlier in the week. Per the MarketBeat summary on Yahoo Finance, Zuckerberg later said the commerce data includes travel purchases.

Plaid is the plumbing. Zuckerberg said, “Plaid told us that they had not seen activity like they are seeing with their connector on any other service before.” TechCrunch describes Plaid as the fintech that links payment apps to bank accounts, so the claim is that Muse users are connecting accounts at an unusual rate. Plaid has not confirmed that publicly in anything I found.

What Meta says it will charge

Zuckerberg on monetization: “For most people, the product should be free, and we will build commerce tools so we get paid a fee based on the transactions.” The MarketBeat summary says the commerce fees would be paid by businesses, alongside subscriptions for advanced models and extra token usage.

Meta gave no rate, no start date and no definition of which transactions count. The transcript, which carries its own accuracy disclaimer, does not use the phrase “take rate” anywhere. Treat the fee as announced intent, not a price list.

Where the agent is headed

Zuckerberg described agents talking to agents. “If you wanna buy something, it is not just navigating to that business’s website,” he said. Li added that “over time, consumer and business agents could connect and transact directly,” and that Meta previewed a personal agent protocol and open standard with partners including Stripe and Shopify. She also cited “at least 500 million self identified small businesses” using Meta’s apps as the audience for business agents, and Superpower Daily’s write-up notes the call gave no fee rate or merchant terms.

Muse is not global yet. Zuckerberg said it is “a compute intensive service to run,” and Meta expects to improve cost efficiency “by more than 10x in the next year.” Fourth-quarter revenue guidance is $73 billion to $76 billion, and full-year capital spending is forecast at $135 billion to $145 billion, per the same MarketBeat summary. Q3 free cash flow was negative $2.1 billion.

Shop Pay Checkout, Known Merchants and the Unpriced Fee

The fee announcement lands on top of a setup that already favors the platforms. Shopify catalogs flow into Muse through Shopify’s catalog access for personal AI agents, and purchases complete through Shop Pay. When I wrote about the checkout integration, no fee structure for agent-routed orders had been disclosed. That is still true today. Meta is saying a fee is coming, not what it is.

Three earlier EP stories matter here. Meta’s agent was reportedly told to rank known merchants first, which tilts the field toward brands with recognition.

A separate report found Muse chat titles exposed in DNS share links, a reminder that this product is young. PayPal also opened checkout to Muse, so Meta is not tied to one payments rail.

The counterpoint

The case for caution is strong, and it comes from Meta’s own wording. Every number above is a CEO statement on an earnings call, not an audited metric. The 25% daily figure covers people who finished setup, and Zuckerberg gave no baseline for how many started and quit. The $200 average includes travel, and the shopping share is a share of users, not of dollars.

Li herself said AI businesses could “begin contributing incremental revenue in 2027,” and she did not break out Muse revenue or costs. Outside Meta, the evidence on AI-driven buying is mixed.

A survey I covered found only 5% buy right after an AI recommendation. A separate study put ChatGPT at 0.62% of Google’s organic traffic.

On the other side, Shopify has said AI-referred sessions convert 80% better. Both can be true: small volume, good conversion.

Sources disagree on one more detail. TechCrunch and the transcript describe 10 million people in the US and Canada after about six weeks, while some summaries drop the geography. The transcript is the better guide, and it says US and Canada.

What a Muse Fee Does to High-Ticket Margins

My read is that this is a small channel with a big structural implication. Ten million setups is real scale for a six-week-old product, but it is a rounding error next to Google Shopping for most stores. The implication is the fee. Once Meta has a take on every order that runs through Muse, the agent has a reason to steer buyers to merchants that pay well, convert well and cause no returns.

Here is the math, and it is hypothetical because Meta has not published a rate. Say your average order is $1,800 and your gross margin after supplier cost is 25%, so $450 of gross profit per order. A 3% channel fee costs $54, or 12% of that profit. At 5% it costs $90, or 20%. At 8% it costs $144, or 32%. Card processing and returns come out of what remains. A high-ticket store running 20% margins would feel an 8% fee as a near wipeout on those orders.

That is why the structure of a high-ticket business matters more than the headline user count. Stores with MAP-protected pricing and 25% to 35% gross margins can absorb a few points. Stores that already fight Google Shopping competitors for a 15% margin cannot. If you want the underlying model, my guide to what high-ticket dropshipping is covers why margin per order is the whole game.

Why Meta needs this fee

Start with the cost side of Meta’s own guidance. Full-year capital spending is forecast at $135 billion to $145 billion, Q3 free cash flow was negative $2.1 billion, and Zuckerberg called Muse compute intensive. He also said monetization would come “somewhat sooner on this than our normal playbook.” A free agent that costs a fortune to serve needs a revenue line, and a cut of every order is the cleanest one available.

The ad business is the other option, and Zuckerberg pointed at it too, describing users who want Muse to improve “the personalization and quality of their content and ads across Meta.” So you may end up paying twice: once in a fee, once in ads that target the same buyer. That is speculation on my part, and Meta has said nothing about bundling the two.

Why high-ticket sellers feel it last

Meta gave no age data for Muse users, so this is instinct, not fact. My experience says high-ticket buyers skew older and still like to call before spending $2,000, which suggests early agent volume will be thin in categories like outdoor power equipment, mobility and home furnishings. Shopping at an average above $200 sounds like apparel, gifts, travel and household goods. The risk for you is not lost sales today. It is a precedent where the agent owns the relationship and the fee scales with order size.

Three scenarios with thresholds

Scenario one: Muse orders stay under 5% of your orders. Ignore the fee debate, keep your feed clean and move on. Scenario two: Muse sits between 5% and 10%. Now the fee is worth a margin review, and you should compare it against your blended ad cost. Meta ad prices already rose, with US and Canada up 23% before Q4, so a transaction fee that replaces ad spend could be a fair trade. Scenario three: Muse passes 10%. At that point it is a primary channel and you negotiate, diversify or reprice.

These are my thresholds, not Meta’s, and they are the same ones I used when Shop Pay checkout arrived. Revisit them when a rate is published.

Your feed becomes the storefront

Li said she expects the market to move away from agents poking through every website and toward “more permissioned structured interactions.” Translate that for a store owner: the agent will stop scraping your pages and start reading a structured feed or talking to your business agent. Whoever has the cleanest data wins the recommendation. Your beautiful category page stops mattering for that buyer, and your spec sheet and warranty fields start mattering a lot.

That also explains the Amazon and Walmart split I wrote about, where one blocks agents and the other welcomes them. My read: Walmart is chasing the order flow, and Amazon is protecting its own storefront. Your catalog already flows out through Shopify, so you sit closer to the Walmart side by default, and your own data and your own pricing rules are the only bargaining chips you hold.

The part nobody is pricing in

A buyer who lets an agent complete a $1,800 purchase has not seen your phone number, your warranty page or your reviews. That raises chargeback risk. Agent buyers who dispute a charge will say they did not choose the product, and your evidence is a log of an AI session. Read my breakdown of chargeback prevention for high-ticket stores now, because the evidence you need changes when no human visited your site. Amex has started to address this with an intent check for AI agent purchases, but my read is that it protects the issuer first and the merchant second.

The operator complexity here is real: feeds, margins, disputes, support and a new fee, all while you try to keep Google Shopping profitable. This is the kind of layered work our turnkey done-for-you build exists to take off your plate.

Want 1-on-1 help deciding how much of your store should be agent-ready, and what your margins can take? Get the coaching details →

Five Moves to Make Before Muse Goes Global

You do not need to rebuild anything. You need visibility, clean data and a margin floor. Here are five things to do this week, and if you want help with any of them you can book a discovery call.

  1. Find your Muse orders. In Shopify, build a saved report on order source and referrer for Meta traffic with unusually fast checkout times, and check it every Monday for the next 30 days. Write down the count and the average order value so you have a baseline before any fee exists.
  2. Audit your product data. Agents read your catalog, not your storefront, so titles, model numbers, specs, price, shipping time and warranty need to be complete. Paste a sample of 20 products into Claude and ask it to flag missing specs. Then compare your results with the checks in the Shopify AI-agent readiness audit, where no store in 122 passed all nine.
  3. Set a margin floor. Calculate gross profit per order after supplier cost and processing, then add a hypothetical 5% channel fee. Any product that falls under your floor gets a MAP-compliant price review, not a discount. Our guide on how to scale a high-ticket store without cutting prices shows the approach.
  4. Capture the buyer after the sale. Agent buyers skip your popups, so build the relationship afterward with a post-purchase flow in Klaviyo (Omnisend works too) that asks for a phone number and offers an accessory or service plan.
  5. Staff for buyers who arrive cold. Put your phone number in the order confirmation and route support through Gorgias. Log call follow-up in HubSpot, and book a discovery call if you want a second set of eyes.

Keep your Google Shopping engine tuned in the meantime, since it still pays the bills. My post on turning Google Shopping clicks into sales for high-ticket products is the place to start. A tool like Semrush shows where your competitors are winning those clicks.

Frequently Asked Questions

Is Muse available outside the US and Canada?
Not yet. Per the earnings call, Meta has not launched it globally and cited compute costs, and the 10 million figure is for the US and Canada only.

Does Meta charge merchants today?
No rate has been published. Zuckerberg said Meta will “get paid a fee based on the transactions,” and the MarketBeat summary says businesses would pay it. Until a number appears, plan with a hypothetical range of 3% to 8%.

Do I have to opt in for Muse to sell my products?
When I covered the Shop Pay checkout news, Shopify catalogs were reported to flow into Muse without a merchant opt-in. Check your Shopify settings for any agent-channel controls before assuming anything.

Does the $200 average purchase apply to high-ticket stores?
No. The figure covers a subset of shoppers and includes travel purchases, so it says little about furniture, powersports or mobility orders in the $1,500-plus range.

Should I move ad budget to Muse?
No, because there is no way to buy placement and no published fee. Keep spending where you can measure return, and watch the retailers choosing to welcome or block AI agents to see how the big players treat this channel.

Which niches suit agent buying best?
Products with clear specs and model numbers, since agents compare on data. Start with my high-ticket niches list. The free version is at ecommerceparadise.com/niches.

Already have a store and want my team to scale it while the channels shift? See the scaling service →

My plan is simple: measure Muse now, protect margin before the fee arrives, and keep owning the customer relationship. Subscribe to the YouTube channel for daily breakdowns. More breaking news coming through the day.

Related Articles

If this was useful, these go deeper:

Share on Facebook
Post on X
Save
Share