Almost every guide to this question is written for somebody who has not started yet. Shopify or Amazon. Your own store or the marketplace. Pick a door, walk through it, good luck.
That is not you. You already have a store, you already have listings, and the marketplace money shows up more reliably than the money from your own site does. Some months it shows up so much more reliably that you have quietly stopped working on the site at all.
So the useful question is not which one you should have picked. It is what the marketplace is doing to your business while it pays you, and how to run both on purpose instead of by drift.
A Marketplace Is a Channel Wearing a Business Costume
Here is the distinction that decides everything else in this article.
An asset keeps producing after you stop feeding it. A channel stops the moment you do. Your customer list is an asset. Your marketplace listing is a channel that pays well enough to look like an asset.
The costume is convincing because the money is real. Orders arrive without you buying a click. Revenue is stable. It has the shape of a business that runs itself. But strip it back and ask what you would still have if the platform closed your account this afternoon, and the answer on the marketplace side is: some inventory in somebody else's warehouse, and a spreadsheet of orders from people you cannot contact.
That is not an argument against selling there. It is an argument for knowing which of the two things you are holding.
This is the same structural trap that catches store owners when they take on their first big wholesale account, and if that is the door in front of you right now, we wrote about that one separately. Different channel, different mechanics, same question underneath: who ends up owning the customer, and what happens to you if they decide they are done.
Count What You Actually Own
Do this literally. Two columns, and go down the list item by item asking whether you have it.
The buyer's email address. Their name and address in your own records. The review they wrote. The ranking that gets you found. The ability to change your price, your title or your photos this afternoon without asking permission. The payment relationship. Whether they could name your brand a week later.
On your own store, you have all of it. On the marketplace, most of that column is empty, and it is empty by design rather than by oversight.
Amazon is the buyer of record, not you. The customer's contact details are withheld, and the messaging system exists so that you can resolve an order, not so you can market to them next month. The reviews you worked for attach to the listing, so they are an asset of that listing rather than of your brand, and they do not follow you anywhere. The ranking that produces your sales is not something you own either. It is something you rent, daily, with performance.
The part owners consistently underrate is brand recall. A shopper who buys from you on a marketplace is not thinking about you. They are thinking about the product and the delivery date. Ask them a month later where it came from and a good share of them will say the marketplace, because as far as their memory is concerned, that is exactly where it came from.
The Fee You See Is Not the Fee You Pay
Sellers compare the monthly subscription. It is the least important number on either side.
On Amazon, the Professional selling plan is $39.99 a month, or you can pay $0.99 per item sold on the Individual plan. Then comes the referral fee, which is a percentage of the total sale price and runs anywhere from 5% to 45% depending on the category. Home and kitchen sits at 15%. Consumer electronics is 8%. Amazon device accessories run 45%. Media items carry a separate closing fee of $1.80 on top. If you use FBA, add a fulfillment fee scaled to size and weight, monthly storage charged by the cubic foot at rates that rise in the fourth quarter, a surcharge on inventory that sits too long, and returns processing in some categories. Then add advertising, because Sponsored Products has quietly stopped being optional for anything competitive.
On Etsy it is $0.20 to list an item, renewed every four months or each time it sells, 6.5% on the total order including shipping, and roughly 3% plus $0.25 to process the payment in the US.
Now price the other side honestly, because the comparison is only useful if you do. Your own store has a platform fee, payment processing in the same neighborhood as everyone else's, an app stack that grows every time you solve a problem with software, and one cost that dwarfs all of them: what you pay to get a stranger to show up at all.
That is the real trade, and it is a trade rather than a saving. The marketplace takes its cut out of your margin, after the sale, on every unit, forever. Your own store takes its cut up front as customer acquisition cost, on the first order only, and then hands you the second and third orders close to free. Which is cheaper depends entirely on whether those later orders actually happen.
If you have never run those numbers per product rather than across the whole business, that is the more urgent job, and it usually rearranges the picture. Blended margin hides the products that are quietly funding the ones that are not.
Rented Land Has a Landlord, and the Lease Changes
Every seller nods along at platform risk in the abstract and then gets surprised by it in the specific. So here are the specifics.
Fees move, and you absorb them. You are not consulted, and there is no version of the conversation where you negotiate.
Rules move too. Etsy's Offsite Ads program is the cleanest illustration in ecommerce: sellers can opt out of it, right up until the shop passes $10,000 in sales over a trailing twelve months. Past that line, enrollment is permanent and the fee is 15% of an attributed order, dropping to 12% once you are above the threshold. Succeed enough and the opt-out disappears. That is a lease term, and you agreed to it.
Rankings move for reasons that have nothing to do with you. A category gets more competitive, an algorithm gets retuned, and a product that carried your quarter is on page three.
And then there is the failure that actually ends businesses. A competitor files a spurious intellectual property complaint, or a metric slips, or something in an automated system decides your account needs review. The account is suspended. What makes this different from a bad month is that it freezes your cash and your inventory in the same moment. The payouts stop, and the stock you already paid for is sitting in a warehouse you cannot direct. Your working capital and your product are both hostage to the same decision, and you are appealing it through a support queue.
Which is the other thing sellers underestimate. When something breaks, you are not making a decision, you are waiting for one. Sellers routinely wait on support to approve a change as small as a product title. Nothing about that is unreasonable at Amazon's scale. It is simply what it means to not own the thing your business runs on.
If this pattern feels familiar, it should. It is the same class of exposure we wrote about in your ad account not really being yours, one platform over.
The Marketplace Is Buying You Something Real. Name It.
None of the above makes marketplaces a mistake, and any article that lands there is selling you something.
Be precise about what you are getting, because it is substantial.
You are getting demand that already exists. Nobody arrives on Amazon to browse a brand. They arrive to buy a thing, with a card saved and an address on file, at the bottom of the funnel. On your own site you have to create that intent and pay for it.
You are getting borrowed trust, and this is the most valuable and least appreciated item on the list. A stranger who has never heard of you will buy the identical product at a similar price on a marketplace and hesitate on your website, because the marketplace is underwriting the risk of the transaction. Returns are known. Delivery is known. Recourse is known. You have not earned that trust and you get to use it anyway. The work of earning it on your own site is real work, and it is mostly done by your existing customers rather than by your copy.
You are getting fulfillment you would struggle to match at your volume, and delivery expectations you could not fund alone.
And you are getting product validation without buying traffic to get it. If a product cannot sell in front of ready buyers who are already looking for it, that is information, and it cost you nothing to learn.
That is a genuinely good deal. It is just a deal for a service, not a business you own.
Run It as Acquisition, Not as Revenue
Here is the part almost nobody writes down, because it is more work than picking a side.
Treat the marketplace as the top of your funnel rather than as the whole thing. That is one sentence and several concrete practices.
Use inserts, and stay inside the rules. A package insert is your one physical touchpoint with a person the platform will not let you email. What Amazon permits is a thank-you, an honest request for a review of any kind, warranty or product registration, support contact details, and genuinely useful material about the thing they bought. What is prohibited is asking for a positive review specifically, offering anything in exchange for one, or screening unhappy buyers away from the review system by routing them to you instead. That last one is where sellers get suspended, and it is not worth it. Register a warranty, offer the care guide, be a brand rather than a box. That is allowed, and it is the part that works.
Build the brand inside the walls. Brand Registry, A+ content and a storefront do not give you the customer, but they do give you the one thing recall depends on, which is somebody noticing your name while they buy. A listing that looks like a brand gets remembered at a rate a bare listing never will.
Decide what you sell where, on purpose. The marketplace is a rough environment for a product that needs explaining, and an excellent one for a product that sells itself on a photo and a price. Your newest work, your bundles, your best-margin items and anything that needs a story are the reasons for someone to come to your own site.
Read it as research. The marketplace will tell you, for free, which variants sell, what people ask in the questions, what the negative reviews of your competitors complain about, and what language converts. That is customer research you would otherwise pay for.
Measure them apart. One blended revenue number is how a business ends up 90% dependent on a channel without ever deciding to be. Track contribution margin by channel and look at them separately.
The Number That Tells You Which Business You Are Really Running
Two figures, and you can pull both this week.
The first is share of revenue by channel. Not a feeling, the actual percentage.
The second is repeat purchase rate by channel. What proportion of last year's buyers bought again this year?
For your own store, that number is knowable. For the marketplace, in most cases, it simply is not, and that is the finding rather than a gap in your reporting. You cannot see whether the same person came back, because you were never told who they were.
Put the two together. If the marketplace is the large majority of your revenue, and your repeat rate there is unknowable, then what you own is not a diversified business. It is one acquisition channel, a good one, that you do not control and cannot audit. Every marketplace order is a first order as far as you can prove.
That is the moment to be honest about which direction you are drifting, because the drift only ever runs one way. The marketplace pays today and the store pays later, so the marketplace wins every single week that you are busy.
The fix is not to leave. For most stores that would be a needless act of self-harm. The fix is to give the owned side a real job, funded and measured, instead of leaving it as the site you will get to eventually. Second orders are where store profit actually lives, and that arithmetic is worth understanding before you decide how much the first order is worth.
Run the marketplace hard. Take the demand, take the trust, take the fulfillment. Then spend the margin it produces on building the thing that keeps producing when the platform changes its mind.
Sorting out which channel deserves which product, and what you can afford to pay for a customer on each, is most of what we do with the ecommerce businesses we work with. If you would rather work it out yourself, the two numbers above are the honest place to start.
