The question isn't "marketplace or own site". It's where you put the next euro, with the catalogue you have today and the cash you have today. And the answer changes depending on the stage each product is in.
Let's get concrete: what each channel gives you, what it takes away, and how to run both instead of picking a side.
What a marketplace actually gives you
A marketplace rents you three things that take years to build:
- Traffic with buying intent. Nobody opens Amazon to browse for fun. They open it to buy. That doesn't happen on your own site unless people already know you.
- Borrowed trust. The shopper doesn't trust you, they trust the returns policy and the channel's customer service. That removes the whole "who even are you?" friction.
- Logistics and payments, solved. With FBA (Fulfilment by Amazon, where they store and ship for you) the operational side is off your desk from day one.
What it doesn't give you, and this is the point: it doesn't give you the customer. No email, no idea what else they bought, no way to contact them when you choose. Amazon gives you a sale, not a relationship. The moment you stop paying for visibility or your ranking slips, the tap closes and nothing has accumulated.
It also sets the rules. Fees change, listing requirements change, review policies change. You adapt. There's no negotiation.
What your own store gives you
Your site gives you margin, data and relationship. There's no referral fee (the cut the marketplace takes from every sale) eating into each order. The exact percentages vary by category and they change, so confirm them in Seller Central before you build a spreadsheet on them.
It gives you data: what people look at before buying, what they abandon in the basket, how often they come back, what they buy after what. That feeds range, bundle and pricing decisions you currently make blind on the marketplace.
And it gives you the email address. A permission-based email list is an asset that doesn't expire when an algorithm changes.
The catch is obvious: you bring the traffic. And it costs. The first sale on your site is expensive; the fifth sale to the same customer is nearly free. The whole own-store model rests on that second part.
Side by side
| Criterion | Marketplace | Own store |
|---|---|---|
| Margin per order | Lower: referral fee, fulfilment and ads | Higher per unit, but traffic is a separate cost |
| Traffic | Comes built in, with buying intent | You generate it from scratch |
| Customer data | Close to none, no usable email | Full: email, history and behaviour |
| Risk | High: someone else's policies, suspensions, algorithm | Less dependency, more execution risk |
| Upfront investment | Low: listings, stock and some ad spend | High: platform, content, traffic and analytics |
| Time to traction | Weeks | Months before volume is steady |
| Repeat purchase | Hard to force, the customer belongs to the channel | The entire reason the channel exists |
Acquisition cost isn't comparable the way you think
The classic mistake is putting "20% ACoS on Amazon" next to "3x ROAS on Meta" and deciding from there. Those aren't the same measurement.
Translate everything into one number: euros of margin left per order after everything. On the marketplace, subtract the referral fee, per-unit fulfilment, returns and advertising. On your own site, subtract the payment gateway, shipping, returns, platform costs and your entire acquisition spend for the month, not just the campaign the tool credited with the sale.
Do it that way and, in the accounts we manage, the first sale on an own site usually costs more than the first sale on a marketplace. The gap opens from the second purchase onwards. If your product is never repurchased, that leverage doesn't exist and the marketplace is very hard to beat.
Watch TACoS (ad spend measured against ALL your sales on the channel, not just ad-attributed ones) to see whether your marketplace brand is healthy or living on spend. On your own site, watch cumulative margin per customer over twelve months, not first-order ROAS.
Working rule: if you don't know what a customer is worth to you over a year, you can't decide what you're allowed to pay to acquire one.
Single-channel dependency is a business risk
This isn't a marketing argument, it's a balance-sheet one. If 90% of your revenue comes through one seller account that can be suspended over a documentation error, a quality complaint or an intellectual property claim, your business has a single point of failure.
Suspensions almost always get resolved. But they can take weeks. Can you go weeks without revenue? That's the real question.
Your own store isn't just a more profitable channel. It's the insurance policy. Even at 15% of your volume, that 15% is what pays suppliers and salaries while you fix the problem.
There's a quieter effect too: if you only sell on a marketplace, your company is worth less the day you want to sell it. A buyer knows they aren't buying customers, they're buying borrowed ranking.
Discover on the marketplace, monetise on your site
This is the sensible way to run both at once.
The marketplace is your lab. You launch there because the traffic already exists, and within weeks you know what sells, at what price, with which main image, and in what words people search for it. Search term reports tell you literally how your customer talks. That's gold for your site copy and for your off-channel campaigns.
Your site is where you squeeze the winner. Once a product has proven demand, you take it to your store with bundles, sizes or formats that don't pay off on the marketplace, with longer content, real cross-selling and an email sequence behind it.
Concrete order of play:
- Launch on the marketplace and let it run for 8 to 12 weeks.
- Identify the two or three products with the best conversion and the best repeat rate.
- Build those product pages on your own site, with original content rather than copied listings.
- Give the buyer a genuine reason to visit your site: usage guide, product registration, extended warranty. Careful here: respect channel policies on buyer communication and traffic diversion. No direct discounts, no review requests.
- Once the customer lands on your site, then and only then, capture the email and work on repeat purchase.
Pricing across channels: don't break it
Common temptation: price lower on your own site to "push" the owned channel. Bad idea, for three reasons.
First, marketplaces monitor prices elsewhere, and a lower price off-channel can affect your eligibility for the Buy Box (the featured buy button on the listing). Check how that policy is worded today in Seller Central, because it has been rewritten more than once.
Second, you train your customer to hunt for the lowest price instead of valuing your brand. Whoever buys for two euros less leaves for two euros less.
Third, you eat your own margin for no reason. Your site's advantage doesn't have to be price: it can be the bundle, the exclusive size, the extended warranty, the gift, early access or a subscription. Differentiate on offer, not on price.
Subscription and repeat purchase: the closing argument for your own store
If your product gets consumed (cosmetics, food, supplements, consumables, spare parts), subscription is the reason your site exists.
A subscriber gives you predictable revenue, amortised acquisition cost and a legitimate reason to talk to them every month. You don't control that on a marketplace: recurring purchase programmes exist, but the customer still belongs to the channel and the discount comes out of your pocket.
Start simple: one product on subscription, a moderate discount, one-click cancellation. Measure average retention at three and six months before building anything complicated. If people leave after the second delivery, the problem is the product or the cadence, not the mechanism.
What to do this week
- Work out real margin in euros per order on each channel, with every cost included. If you don't have that number cleanly, that's job one, and no decision is possible without it.
- Check what share of revenue depends on a single channel. If it's above 80%, set a twelve-month reduction target and write it down.
- Pick one proven marketplace winner and build it properly on your own site: original content and a different offer (bundle, size or warranty).
- Check you aren't undercutting your marketplace price on your own site without having decided that on purpose.
- If your product gets repurchased, design a subscription test with a single product and let it run three months before judging it.