How to sell on Meta Muse
Short answer for most shops: you already do.
Meta launched Muse on 8 September. It's an assistant that goes off and does errands for people, including buying things. Within hours of launch, Shopify said its merchants' products were already inside it, checking out through Shop Pay. Nobody signed up for that. There was no waitlist and no developer account, and if your products sync to Shopify's catalog, Muse can find them right now, today, while you read this.
So the interesting question isn't how to get in. It's why it might look straight past you.
What agentic commerce actually changes
The phrase for this is agentic commerce: software doing the shopping instead of the shopper. Agentic commerce sounds like a payments problem and mostly isn't one. For a shop like yours, agentic commerce is a question of whether Muse can read your product pages at all.
Robots don't squint
When a person lands on a product page and can't find the shipping cost, they don't leave. They poke around. They check the footer, they add it to the cart to see what happens, they assume it's probably a fiver and carry on. If your returns policy is a JPEG, they read the JPEG. If your description says "premium materials" and nothing else, they look at the photo and decide for themselves.
People patch the holes without noticing they're doing it. It's why you can run a shop for years with half the details missing and never feel the cost.
A robot won't patch anything. It's filling in a comparison table across five shops, and a cell it can't fill is a cell it leaves empty. Empty usually loses.
There's a version of your product page you've never looked at. Underneath the design, there's meant to be a plain block of data saying what the thing is, what it costs, whether it's in stock. Most shops never built that part, because until about three weeks ago nothing was reading it.
What we keep finding
We scan ecommerce sites for a living. Across the last 220:
- 48% had broken or missing product data, the machine-readable kind
- 48% had nothing saying who the company even is
- 50% didn't show shipping cost before checkout
- 36% had a returns policy that was incomplete or buried
- 69% had reviews no machine could read
Just over half failed at least one of those outright.
That reviews number is the one I'd stare at. Most of those shops have plenty of reviews. Good ones. They're sat there on the page in a widget that loads after everything else, and as far as a robot's concerned they don't exist. You did the hard part, getting customers to write them, and then lost the benefit at the last step.
Four things, cheapest first
Start by searching for your own product the way a customer would actually say it. Not your brand name, which proves nothing. Say what the thing is: "waterproof dog coat for a whippet". If three competitors come back and you don't, that's readable-ness, not Muse.
Then add the tag that says who you are. It's a small standard snippet with your business name, logo and contact details, and half the shops we scan don't have one. If a robot can't confirm you're a real company, recommending you is a risk it doesn't have to take. Ten minutes, and it's the cheapest thing on this page.
Then write the boring facts as words. Shipping price, delivery window, returns window, sizing, materials. Actual text on the page. Not inside an image, not in a PDF, not behind a tab that only loads when someone clicks it. These are exactly the questions the robot has to answer before it'll put you forward.
And get the reviews into a form a machine can see. Most review apps can do this. Most shops never switched it on.
The thing you can ignore
There's a lot of noise right now about AI payment systems, mostly about something called MPP. It's real. It's probably not yours.
MPP is for robots buying things for themselves: computing power, data, an API call they need to finish a job. If you sell coats to humans, that's not your situation. Your customer is still a person, the robot's just running their errand, and the money goes through Shop Pay, which works already.
I'd rather you spent the afternoon on your Organization tag.
Where these numbers come from
Every percentage on this page comes from our own scanner, run against the last 220 ecommerce sites we audited before 2026-09-28. It's an automated pass over live pages: what loads, what's in the markup, what a crawler can and can't reach. Not a survey, not an estimate, and not a sample we picked to make a point. Where a check didn't apply to a site, that site isn't counted in its denominator, which is why the totals differ per row.
We'll re-run it as these assistants change. If a number here is stale by the time you read it, tell us and we'll fix it.
What I don't know
Muse is three weeks old. It did 2.5 million downloads in a fortnight, quicker than ChatGPT managed, and Amazon has already blocked it from their site. How it decides what to recommend will change, probably more than once, and anyone telling you they've worked out its ranking rules this month is guessing.
The floor won't move, though. Every one of these assistants has to read your shop before it can talk about it. A shop that machines can read will work with whichever one wins, and the same work makes you more legible to Google and to your own site search. It's the rare bit of 2026 hype where the prep work is useful even if the hype evaporates.
If you're on Shopify
Most of the above is a checklist you can work through by hand. We also built it into a Shopify app, mostly because we got tired of sending people the checklist.
Mimetic Conversion & AEO Agent is free to install. It scans your live store and tells you which of these a robot currently can't read on your actual products. AEO stands for answer engine optimisation, which is a clumsy name for exactly the problem this page is about: being legible to the things that answer questions on your customers' behalf. Where a fix is safe to make in your theme, it'll make it and let you roll it back.
Or talk it through with us. If you'd rather not guess at which of these applies to you, grab 30 minutes. We'll pull your shop up, look at what a robot can and can't read on it, and tell you which fixes are worth your time. Some shops need one. Some need none, and we'll say so.