In short
- You open a marketplace to escape advertising dependency. It brings demand that does not come from your campaigns, so it does not stop when the cost per click rises.
- Catalogue mapping decides everything. It is the most thankless work, it takes weeks, and no tool does it properly for you.
- The price you send is not your price. Tax inclusive, VAT of the delivery country, local currency. One field filled wrongly and you sell at a loss without seeing it.
- Do not confuse a feed manager with a publishing provider. The first synchronises your data and leaves you in control; it costs money and saves time. The second publishes on your behalf, directly with the marketplace, and that is what takes control away.
- Delegating publishing means delegating what actually sells. Product merchandising and promotional campaigns are not technical settings, they are commercial decisions.
- Losing control of price damages your distribution network. A promotion decided without you undercuts your price in front of your own retailers. That is no longer a margin question, it is a commercial policy question.
- Allow six to eight weeks between signing and the first product going live. Content and translations, not the technology, take up most of it.
- Our advice: run your marketplaces yourself. One, mastered end to end, returns more than five delegated.
A brand opens a marketplace. The catalogue goes out, the first orders arrive, everyone is pleased. Three months later the channel’s margin is below the site’s, nobody knows why, and the best references are invisible in the platform’s results.
This is the most common scenario we meet, and it almost never has a technical cause. It comes from three decisions taken too quickly: what goes into the catalogue, at what price it is sent, and who keeps their hand on publishing.
Why a brand opens a marketplace
The first reason is not volume, it is dependency. A brand whose acquisition rests almost entirely on online advertising is exposed to a decision it does not take: a rise in cost per click, an algorithm change, a suspended account, and revenue stops dead. The channel is not theirs, they rent it.
A marketplace brings customers who do not come from your campaigns. It has its own traffic, its own buying habits, and it discovers you through a product search, not through an advert. It is not one more channel in the quantitative sense, it is a source of demand that does not depend on the same tap.
That is also why it deserves real investment rather than an automatic feed. Diversifying your distribution channels only matters if you keep control of the ones you open. Otherwise you swap one dependency for another, and lose control of your prices into the bargain.
A marketplace is not one more channel
It is a second trade, with its own rules. On your site, a listing error costs a little conversion. On a marketplace it costs a suspension, then a delisting, and you do not win back a lost position by fixing it the next day.
The difference lies in who arbitrates. On your site, you decide what is promoted, when, with what offer. On a marketplace, the platform’s algorithm decides, from your data and your performance. Your only lever is the quality of what you give it: complete listings, coherent prices, accurate stock, delivery times kept.
That is exactly the subject of our e-commerce and marketplaces expertise, which also covers choosing your platform between PrestaShop, WooCommerce, Shopify and Shopware.
Catalogue mapping: the work nobody wants to do
Every platform imposes its own taxonomy. Your sizes are not its sizes, your colours are not its colours, your categories do not exist there. A dress in five sizes and three colours becomes fifteen references, each with its identifier, availability and price.
The work is to translate, attribute by attribute, your model into theirs. It is long, it is tedious, and it decides everything else: a badly mapped listing is rejected on import, or worse, published with the wrong attribute. A garment filed in the wrong category will never appear in results, whatever its price.
The mistake to avoid Pushing the whole catalogue “to see what works”. A marketplace judges a seller on the average quality of their listings and on their cancellation rate. Fifty well-documented references, in stock, with real photographs and delivery times kept, build a seller reputation. Three thousand approximate ones destroy it, and that reputation then governs your visibility across the whole catalogue.
Price arbitrage: the margin is lost in one field
This is where we see the most money disappear, without anyone noticing before the year-end review.
A marketplace’s price field almost always expects a tax-inclusive amount, with the VAT of the delivery country, in the local currency. Your back office, meanwhile, usually reasons net, in your own currency. Sending one into the other produces no error: the listing publishes, it sells, and every order cuts into a margin nobody decided to give up.
| What changes from one country to the next | What happens if you ignore it | How to handle it |
|---|---|---|
| The VAT rate of the delivery country | The same displayed price returns less in a high-rate country, and margin varies for no apparent reason | A price calculated per country, never one price converted |
| The local currency | The exchange rate moves, the displayed price does not. Margin degrades slowly and nobody watches | A conversion rule reviewed at fixed intervals, with a margin floor |
| The platform’s commission | It applies to the tax-inclusive price, so to a higher amount than people calculate | A selling price built from the target margin, not from the site price |
The habit to acquire: your marketplace price is not derived from your site price. It is built backwards, starting from the margin you want to keep after commission, VAT and delivery costs.
The real problem: those who promise to publish everywhere for you
First, a distinction the market is happy to keep blurred, and which changes the subject entirely.
On one side, feed managers. These are tools: they take your catalogue, transform it into the format each marketplace expects, and synchronise stock and prices. The seller account stays yours, you keep your hand on publishing, merchandising and promotions. They take nothing away from you. Their issue lies elsewhere, in the budget: the bill is a monthly subscription that grows with every channel added, and it comes as a quote rather than a public price list, since vendors in this market rarely publish one. In exchange, the time saved is real, and across several channels they become hard to do without.
On the other side, catalogue publishing providers, who deal directly with the marketplaces and either sell under their own account or run yours. That is no longer a tool, it is a commercial intermediary. And what you delegate to them is not technology, it is commerce.
- Control of publishing. You no longer decide which reference goes out, when, or in what form. A listing rejected by the platform often stays rejected for weeks, because nobody watches the import responses line by line.
- Control of merchandising. Every marketplace has its visibility mechanisms: enriched content, position within a category, eligibility for an editorial selection. Those levers require going into the seller interface. A provider publishing at volume does not pull them, and a feed pulls them no better.
- Control of promotional campaigns. This is the most expensive one. A platform’s commercial operations are prepared, applied for, stocked and steered day by day. An automatically distributed catalogue misses all of them.
The result is presence without performance: you are listed everywhere and selling nowhere. And since the channel does not take off, people often conclude that the marketplace “does not work for us”, when nobody has actually worked it.
What really makes brands walk away: price in front of your network
There is something worse than channel performance, and it is the reason we ended this kind of arrangement. When the intermediary decides promotions alone, your product appears discounted on a marketplace while your retailers sell it at full price. You did not decide it, and you often learn it from them.
For a brand that also distributes through a network, this is no longer a margin question, it is a question of price coherence. A retailer who finds your reference cheaper elsewhere, with no explanation and no correction possible, stops stocking you. And experience shows that obtaining a price correction from an intermediary running dozens of marketplaces is close to impossible: price is a parameter of their system, not of your commercial policy.
That is the criterion that should decide before all others. If you cannot change a price within the day, and prove that it changed, you are not in control of your distribution.
Our position, and it is a firm one Run your marketplaces yourself, with your hand on publishing, merchandising and promotions. One marketplace mastered end to end returns more than five delegated to an automatic feed. A feed manager, for its part, has a real use: keeping stock and prices up to date as channels multiply. It takes nothing away from you, it costs you. What should be refused is handing over publishing itself to an intermediary who will sell under their own account and decide your promotions.
The method: four steps before the first order
- Choose what goes out, and what does not. Not the whole catalogue. The references where you keep your margin after commission, where stock is reliable, and where the listing is already complete. Thirty to fifty references are enough to learn.
- Map, then have it reviewed. Attributes, sizes, categories, images in the required formats. Have the mapping checked by someone who knows the product, not only by whoever knows the file.
- Build prices backwards. Target margin, then commission, then the country’s VAT, then currency. The selling price is the result, never the starting point.
- Put monitoring in place before opening. A marketplace integration degrades in silence: a feed failing overnight warns nobody, stock freezes and the platform keeps selling. Daily reconciliation between what the platform displays and what your back office delivered is the minimum.
An immediate self-diagnosis, with no tool: take ten references already listed on a marketplace and recalculate by hand the margin actually collected after commission, country VAT and delivery costs. Compare it to the margin on the same product sold on your site. If you cannot do that calculation, that is already the answer.
How long it takes, and what takes that long
From signing to the first product going live, we observe six to eight weeks. It is neither a sales figure nor a safety margin: it is what the work demands when it is done properly, and there is no accelerated version that holds.
| Period | What happens | Who does it |
|---|---|---|
| Weeks 1 and 2 | Framing: which references go out, price arbitrage per country, validating the delivery promise | You, with us. It is a commercial decision, not a technical one |
| Weeks 2 to 5 | Catalogue mapping and content production: rewritten descriptions, translations, images in the required formats | The longest part, and the one that decides visibility |
| Weeks 4 to 6 | Technical connection, end-to-end tests, a trial run on a few references | Us. The shortest part of the project |
| Weeks 6 to 8 | Progressive publishing, platform validation, corrections, first orders | The platform has its own validation pace, which cannot be accelerated |
Talking to the platform is itself a source of delay
One thing nobody anticipates and which weighs on the schedule: how you communicate with the marketplace.
On many marketplaces, most of the relationship goes through the seller portal’s internal messaging. No phone, no named contact at the start: you ask a question about a rejected attribute, a format rule, a price correction, and the answer arrives when it arrives. Each exchange takes days, sometimes more depending on the season.
This is not a flaw in the platform, it is how it works, and the project has to be built with it rather than against it. Three habits avoid losing weeks.
- Group your questions. One message asking eight precise questions gets eight answers. Eight separate messages get one answer a week.
- Never block on a pending answer. The project must advance in parallel, on content and on prices, while a question sleeps. A sequential plan where each step waits for the previous one stretches mechanically.
- Record everything. Answers arrive in a thread, often from different people. Without a decision log, the same question comes back three months later and you start another round trip.
It is one of the reasons our six to eight weeks do not compress, even with a perfectly organised client. Part of the delay is not yours.
Content: do not copy your own listings
This is the blind spot of almost every project. The temptation is to reuse your site descriptions as they are, since they already exist. It costs twice.
First with search engines. Your product page and the marketplace listing then carry the same text, on two domains, one of which has far more authority than yours. The common outcome is that the platform’s page ranks ahead of yours for the name of your own product: you pay a commission on a sale you would have made directly.
Then with the platform itself. Many reject or downgrade listings whose text is identical to another seller’s, which happens mechanically when several resellers reuse the manufacturer’s description. A description specific to your listing is a visibility factor, not a nicety.
And you have to count the languages. A European marketplace expects content in the language of each country served, not a machine translation left as is: badly translated technical attributes trigger product returns, and returns weigh on your seller rating. It is editorial work, to be budgeted as such from the framing stage.
What will certainly stretch the timeline Discovering in week 5 that descriptions have to be rewritten and translated, when the budget only covered the technical connection. Content is decided in week 1, together with prices per country. They are the only two subjects that genuinely block a launch, and the two that projects underestimate.
The four approaches, and what they are worth
| Approach | What it gives you | Who it suits |
|---|---|---|
| The seller interface, by hand | Total control of publishing and promotions. But manual entry does not hold beyond a few dozen references, and stock drifts out of sync | A first channel, a small catalogue, a learning phase |
| The feed manager | Real time saved on synchronising catalogue, stock and prices. The seller account stays yours, and so does control. The issue is cost: a monthly subscription that grows with each channel, quoted rather than published | Several channels already open, when manual synchronisation no longer holds |
| A direct connection, built for you | Control over everything, mapping included, with automatic synchronisation. Requires an initial investment and someone following it | A channel that genuinely counts in revenue |
| The provider who publishes for you | Nothing left for you to do. But you lose fine-grained publishing, merchandising and campaigns, and obtaining a price correction becomes uncertain | No case where we recommend it to a brand that cares about its distribution |
Tooling: what you buy and what gets built
The distinction that matters is not between in-house and bought, it is between a connector dedicated to one single marketplace and a tool claiming to cover them all. The first follows one platform’s rules: its attributes, its formats, its merchandising mechanisms, its commercial operations. It exposes those levers to you instead of hiding them. The second has to find the lowest common denominator across fifty platforms, and that denominator excludes precisely what makes one platform different from another.
A dedicated connector can perfectly well be a product, and that is what we industrialise platform by platform. This know-how is not theoretical. We have built these integrations in production, on European marketplaces, with what they genuinely demand: the catalogue mapped variant by variant, prices recalculated per country tax inclusive and in local currency, stock held without overselling, shipments and returns brought back into the back office, and the daily reconciliation that catches a feed that fell over during the night.
What stays bespoke is your catalogue mapping and your pricing rules: they depend on your references and your commercial policy, not on the platform. The rest can be industrialised.
Around the connector, the pieces that hold the channel together can be bought, and they weigh heavily. We deploy for our clients the modules published by Datafirefly Limited, the sister company of our agency: multi-warehouse stock routing so you do not oversell across channels, order export to the logistics provider with tracking so delivery times are kept, and on listing compliance the product safety module along with the price reduction one, compliance having become the leading reason for blocking before publication.
The method comes before the tool, and here more than anywhere. No module decides for you which references go out or at what price. Both of those are commercial calls, they belong to you, and they are what makes the channel’s margin.
A word on a neighbouring workstream, because it lands at the same moment: selling to businesses on a marketplace puts you in front of the French e-invoicing reform, whose first deadline applies to every company in France from September 2026. You may as well handle both in the same data project.
What this says about something bigger
The temptation of maximum coverage runs through all of digital marketing: be everywhere, straight away, with as little effort as possible. Marketplaces offer a useful rebuttal, because the penalty there is immediate and measurable. A seller who pushes everything and holds nothing watches their rating fall, their visibility disappear and their channel die.
Our conviction fits in one sentence: on a marketplace, control beats coverage. One marketplace run seriously teaches you what your organisation can actually carry, and that lesson pays forward into the second.
FAQ
Should we open several marketplaces at once?
No. Catalogue, logistics and compliance constraints differ from one platform to the next, and it is the first one that teaches you what your organisation can carry. One, run well, teaches more than three skimmed over, and the lesson pays forward.
What is the difference between a feed manager and a publishing provider?
A feed manager is a tool: it transforms and synchronises your catalogue, stock and prices, but the seller account stays yours and you keep your hand on publishing, merchandising and promotions. A publishing provider is a commercial intermediary, dealing directly with the marketplace: it decides what goes out and often what is discounted. The two are sold using similar words, and they commit you to completely different things.
Is a feed manager worth its price?
On a single channel, rarely: synchronisation holds without it. From three or four channels the time saved becomes hard to replace. What to watch is the cost trajectory: the bill is a monthly subscription that grows with each channel added, and vendors quote it rather than publish a price list. Compare it to the margin the channels actually generate, not to their revenue.
Why does price control matter more than the channel’s margin?
Because a brand that also distributes through a network puts its retailer relationships on the line. A promotion decided without you shows your product discounted while your retailer sells it at full price. The retailer does not argue about the channel’s margin: they stop stocking you. The test is simple: if you cannot change a price within the day and prove it changed, you are not in control of your distribution.
Does a marketplace really reduce dependency on advertising?
Yes, provided you work it. It brings demand from its own traffic, which does not stop when the cost per click rises or an advertising account is suspended. But diversification only matters if you keep control of the channel you open: otherwise you swap one dependency for another, and lose control of your prices as well.
How long between the decision and the first sale?
We observe six to eight weeks between signing and the first product going live. The technical connection is the shortest part: catalogue mapping, rewriting descriptions, translations and price arbitrage per country take up most of it, plus the platform’s own validation pace.
Can we reuse the descriptions from our own site?
It is inadvisable, for two reasons. The marketplace page, carried by a stronger domain, often ranks ahead of yours for the name of your own product, and you pay a commission on a sale you would have made directly. And many platforms downgrade listings whose text is identical to another seller’s, which happens as soon as several resellers reuse the manufacturer’s description.
How many references should we send to start?
Thirty to fifty are enough. Choose the ones where margin holds after commission, where stock is reliable and where the listing is already complete. The platform judges your seller reputation on average quality and cancellation rate, not on the volume sent.
Can Dotsland help?
Yes. Framing what goes out and at what price, catalogue mapping, the technical connection, stock rules, and the daily reconciliation that avoids suspension. It is the heart of our e-commerce and marketplaces expertise. Let’s talk, or start with the self-diagnosis above: recalculating the real margin on ten references is often a wake-up call.