Marketplaces versus your own site

Every seller who has been on a marketplace long enough has the same conversation with themselves. Revenue is up, the account is in good standing, and then a policy notice arrives — a category deprioritised, a new mandatory ad placement, a fee schedule quietly revised — and the seller realises they cannot say, with any precision, what would happen if the account were suspended tomorrow. Not "revenue would drop." Something closer to: they do not know who half their customers are. The platform does.

That is worth sitting with before making the case against marketplaces, because the honest starting point is conceding what they do better than almost anything a small seller can build alone. A new business with no search history and no audience has two problems before a single sale happens: nobody can find them, and nobody has a reason to trust them yet. Ranking on your own domain for anything commercial takes a long runway — the kind covered in acquisition channels that survive scrutiny — and trust, for a seller with no track record, has to be borrowed before it can be earned. A marketplace solves both on day one, for a fee, and for a genuinely new seller that fee is often the cheapest path to a first hundred customers there is. The real question is not whether to use one, but how long it should stay the main channel rather than a supplement.

The fee you can see is not the fee you pay

Start with the number that is actually printed, because even that one understates the cost. A marketplace's headline commission — ten percent, fifteen, whatever line appears on the payout statement — rarely describes what selling on the platform actually costs. Layered on top sit payment processing fees, listing fees, fulfilment or storage charges where the platform also warehouses inventory, and, increasingly, placement in search results that has quietly stopped being free. What used to be organic ranking inside the marketplace's own search has become, on platform after platform, an auction a seller can choose not to enter, at the cost of sinking below competitors who did. Nobody calls this a second commission. It functions as one, and a seller budgeting off the headline rate is budgeting off a number the platform itself no longer treats as complete.

This is not a marketplace-specific pathology — every channel has a true cost above its advertised one, and paid social carries the same dynamic, covered at length in when paid social works for B2B. What differs is that the marketplace version compounds without a ceiling: the fee scales with revenue indefinitely, with no point at which a seller graduates off it the way a ranked page or an owned email list is eventually paid off outright. Selling more does not reduce the percentage. It just produces more of it.

The building on the marketplace's side

Before the seller's own presence enters the picture, it is worth being precise about what it needs to do, because the mistake is assuming it means matching the marketplace feature for feature. It usually does not. A freelancer or consultant sending a marketplace buyer somewhere direct is not trying to replicate a storefront — they need one page that proves they are a real, findable person independent of the platform, with a way to be contacted.

That is a narrow job, and it is worth naming the tool built for exactly that scope. reach turns a CV into a one-page site: upload a résumé and a photo, answer a short form, pick a look, and the page is generated in about twenty seconds, live at a free subdomain in under two minutes. For a freelancer who has been living entirely inside a platform like a freelance marketplace, that is the entire cost of having something to point a client toward once the platform relationship has run its course — no build project, no empty template. Premium, needed for a custom domain, is $4.99 a month or $49 a year. Prices checked August 2026.

It will not replace a marketplace listing for anyone running an actual storefront: reach has no e-commerce of any kind — no products, no cart, no checkout — and no contact form, just a mail link. That is exactly why it fits the narrow job above and not a larger one. A seller building the direct relationship still closes it by email or a call, not by processing a transaction on their own domain; anyone who needs an off-platform storefront is looking at a heavier build, and the marketplace stays the transactional layer regardless.

Who has the customer's name is the entire question

The deeper cost sits one layer below the fee schedule. On most marketplaces, the platform holds the buyer's contact details, order history and communication thread, and the seller sees a masked or restricted version — enough to fulfil an order, rarely enough to reach that person again without going back through the platform. That is not an oversight: if sellers could freely export buyers into their own list, the marketplace would be disintermediating itself every time a repeat purchase happened.

The consequence lands at the exact moment it hurts most — renewal, or the second purchase. A direct customer who bought from a seller's own site left an email address the seller owns; a marketplace customer is, from the seller's side, a stranger again the moment the transaction closes, findable only if they search for the same listing a second time. The seller did the work of winning them once. The platform decides whether that happens again, and has no structural incentive to make it easy — every repeat purchase it keeps inside its own recommendation loop is one it keeps earning commission on rather than one it hands over for free.

This is the sense in which the customer relationship, not the commission, is the real price. Commission is money; the customer record is optionality — the ability to reach someone again on your own terms, without asking. A seller who has sold to thousands through a marketplace and owns none of their contact information has built revenue without building an asset, and the gap becomes visible the moment the platform changes a fee, an algorithm, or a policy, and there is no list to fall back on.

The concentration threshold

This is where the analysis stops being about tactics and starts being a business-continuity question. There is no verified percentage that applies uniformly across categories and company sizes — anyone quoting one is guessing. The useful test is survivability: if this platform suspended the account tomorrow, for a policy violation nobody knew they'd committed or a category it decided to deprioritise, would the business survive the quarter?

For some sellers the honest answer is no, and that deserves attention rather than rationalising away, because a channel with no fallback is a single point of failure wearing a channel's clothing. Concentration often looks like success right up until it doesn't — revenue growing becomes a reason to lean in further rather than diversify, exactly the moment the risk compounds fastest, since the revenue base a suspension would threaten next year is larger than this year's. The practical marker is not a percentage of current revenue. It is whether a second channel could absorb a meaningful share of it within a sales cycle if it had to. If no other channel currently produces enough to matter, the marketplace is not diversified risk — it is the whole business, with a platform's name on the lease.

Moving buyers off-platform, inside the rules

Most marketplace terms of service explicitly prohibit soliciting a buyer found through the platform to transact directly — a discount for paying outside it, a business card asking for a direct order next time, contact information inserted where the platform expects packing details only. Enforcement ranges from a warning to account termination, the exact scenario the previous section described. Treating that rule as an obstacle to route around is the wrong instinct; it is closer to a boundary defining what the legitimate version of this strategy looks like.

The legitimate version does not intercept the marketplace transaction. It builds a reason for the buyer to volunteer into a direct relationship after the sale has closed on the platform's own terms — a note that points to useful content rather than a discount, a warranty registration that happens to collect an email. None of that solicits the sale away from the platform; all of it earns a second touchpoint the buyer chooses to give. The distinction the terms of service actually care about is inducement versus invitation, and platforms can tell the difference between a seller building a relationship and one trying to take the cut off the table — treating the listing as the top of a funnel rather than the whole funnel, and converting a fraction of buyers into people who come back next time unprompted, because they remembered the name rather than the platform. That fraction compounds slowly and rarely gets celebrated in a monthly report, but it is the only lever that reduces concentration without reducing revenue.

The verdict, by stage

As a first channel, for a genuinely new seller with no audience and no search authority, a marketplace is frequently the right call — it solves findability and trust simultaneously at a price that usually beats building both from nothing. Treat that period explicitly as a bootstrap with an exit condition, not an arrangement to settle into.

As a supplementary channel alongside a direct site that already earns its own customers, a marketplace is close to free money. It reaches buyers who would never have searched by name, at a fee that is easy to accept because it is not the only channel funding the business, and the concentration risk that makes it dangerous elsewhere simply doesn't apply when it is one revenue source among several credible ones. This is the healthiest version of the relationship, and the one most sellers should be aiming toward.

As the only channel for an established business, it is a trap regardless of how the trend line looks, because the growth being watched is growth in someone else's asset, priced in revenue whose future terms nobody but the platform controls. The sellers who get hurt by this are rarely the ones who never tried a marketplace — they are the ones who tried one, it worked extremely well, and nobody came back to ask what would happen if it stopped. A services business making the same mistake with its own site, building traffic that never converts into an owned relationship, runs into a parallel version of this problem, which is why the services page that has to rank and sell treats conversion into a direct relationship as the point of the page. Any channel that produces revenue without producing a relationship you own is rented, and rent is fine to pay as long as you know what happens the month the landlord changes the terms.