Marketplace Take Rate: How Much Commission to Charge and How to Defend It
What a marketplace take rate is made of, what real platforms charge, and how to work out your own ceiling from one seller order.
A take rate is the platform's share of transaction value counted across every stream at once: sales commission, listing fees, seller subscriptions, paid placements and the margin on payments and logistics. Public filings show a spread from roughly 12 percent of GMV on a large product marketplace to nearly 28 percent on a services one. Your own number is not somewhere in that spread, though, because two other figures decide it: the seller's contribution margin and what the same order would cost them without you.
Key takeaways
- A take rate is the sum of every stream, not the headline commission. Allegro reported 12.26 percent of GMV in Poland for Q4 2025, Etsy 24.2 percent for full-year 2025 and Fiverr 27.7 percent, and each of those numbers covers a different set of fees.
- The ceiling is set by the seller's contribution margin and by what the same order would cost them elsewhere, not by an industry average. If a seller can buy that demand more cheaply somewhere else, the benchmark stops mattering.
- Mature platforms raise their take rate in layers rather than in percentage points: advertising, logistics and financing grow while the base commission stays put. At Allegro, advertising alone is 2.2 percent of GMV.
- Disintermediation risk grows with the trust you build yourself. Falling commissions on repeat transactions and services that cannot be recreated off-platform work better than a clause in the terms.
What follows covers five things in order: what a take rate is made of, what levels company filings and public price lists actually confirm, how to calculate your own ceiling from a single order, how to spot a rate that is set wrong, and what to do about sellers and buyers settling deals off-platform. Every figure comes with a source and a date, because price lists change mid-year.
What a take rate is actually made of
Operators designing monetisation from scratch tend to think in one number: a percentage per transaction. Mature platforms capture value in several layers at once, and each layer has its own price sensitivity and its own moment when the seller feels it.
- Transaction commission: a percentage of order value, usually varying by category, because categories carry very different margins.
- Listing fees and seller subscriptions: a fixed amount per listing or a monthly access fee. Paid regardless of sales, so they filter out casual sellers while deterring the ones still testing the channel.
- Margin on payments: the gap between what you charge for processing and what you pay your provider. Stripe's Polish price list quotes 1.5 percent plus PLN 1.00 for standard EEA cards and 3.25 percent plus PLN 1.00 for international ones (accessed 7 August 2026), so your margin only starts above those rates.
- Paid placements and advertising: sellers buying visibility inside demand you already generated. Usually the fastest-growing layer.
- Services: logistics, storage, returns, financing. A separate product on paper, indistinguishable from commission once the seller totals up the month.
The distinction that saves seller conversations: the base commission is your price list, the take rate is the invoice a seller sees at month end. They negotiate the first number, but they decide about the channel on the second.
What others charge: numbers from filings and price lists
The figures below come from company results and public price lists. They are here to show the spread and how much the definitions differ, not to be copied. A take rate reported by a listed company usually covers the whole monetisation stack, while a published rate covers a single layer.
| Platform and model | What sits in the bill | Level (as of 7 August 2026) |
|---|---|---|
| Allegro, product, Poland | Commission, advertising, logistics and financial services | 12.26 percent of GMV take rate in Poland in Q4 2025 (+0.25pp YoY); advertising alone is 2.2 percent of GMV |
| Etsy, product, global | Listing fee, commission, payments, on-site ads | 24.2 percent of revenue to GMS for 2025; 24.5 percent in Q4 |
| Amazon, product, global | Category referral fee, subscription, fulfilment and ads on top | Referral fee typically 8-15 percent (minimum USD 0.30), Professional plan USD 39.99 per month |
| Fiverr, services | Seller commission plus buyer service fee | 27.7 percent marketplace take rate for 2025 (+10 basis points YoY) |
| Upwork, services | Freelancer service fee plus client fees | USD 683m marketplace revenue on USD 4.03bn GSV in 2025, roughly 17 percent |
| Airbnb, services | Host fee and guest fee, or a host-only fee | 3 percent from hosts plus 14.1-16.5 percent from guests; under host-only, most hosts pay 15.5 percent |
| Faire, B2B wholesale | Order commission plus a customer acquisition fee | 15 percent on repeat marketplace orders, 25 percent on a new customer's first order, 0 percent on Faire Direct |
Sources: Allegro Q4 2025 results presentation, full-year 2025 releases from Etsy, Fiverr and Upwork, sell.amazon.com pricing, Airbnb and Faire help centres. Each figure covers a different set of fees, so they are not directly comparable.
Three caveats. First, everyone measures differently: Etsy divides revenue by GMS, Fiverr divides marketplace revenue by GMV, Allegro reports a separate take rate for Poland. Second, published rates exclude costs that still hit the seller, fulfilment above all. Third, these are the numbers of platforms with years of liquidity behind them, not a starting point for a new one.
The size of that gap shows up in a Marketplace Pulse estimate from 13 February 2023: a typical FBA seller hands Amazon more than half of their revenue once you stack a 15 percent referral fee, 20-35 percent in fulfilment costs and up to 15 percent in advertising. That is an analysis of a sample of seller income statements rather than Amazon data, so treat it as an order of magnitude. The lesson holds anyway: the base commission is often the smaller part of the bill.
B2B marketplaces: there is no benchmark here
For B2B platforms we found no public, comparable take rate data. Listed companies in the segment do not report it in that form, and the round-ups circulating online disclose neither methodology nor sample. The only hard numbers are individual operators' price lists, Faire being the clearest. If someone quotes you an average B2B commission, ask for the source before you build a model on it.
In practice, pricing logic on a B2B marketplace sits closer to contract price lists than to a single rate for everyone: different volumes, different payment terms, different conditions per account. A flat commission in wholesale usually means your largest customers pay the most, which is precisely the opposite of how that market works.
Why a benchmark is not the answer
A market average says nothing about your category. On an EUR 20 basket, a 15 percent commission is EUR 3, less than the cost of handling a single dispute. On an EUR 1,000 basket, the same rate is EUR 150 and the seller will start pricing what it would take to win that customer alone. One number, two different worlds.
Calculating your own take rate: the seller sets the ceiling
The method we use when designing monetisation starts not with your revenue plan but with the economics of a single order at a typical seller. Without that number, a price list is a wish.
- Step 1. Work out the seller's contribution margin on one order: order value minus cost of goods, packaging, the shipping they carry and a provision for returns.
- Step 2. Subtract the profit they will not go below. What remains is the theoretical ceiling: the most the channel could ever take.
- Step 3. Price the alternative: what the same order costs the seller without you, counting advertising, payments and running their own store.
- Step 4. Take the lower of the two. The real ceiling is the alternative, not the margin. A commission above the alternative holds only while the seller cannot execute it.
- Step 5. Subtract your own variable costs: payments, disputes, support, moderation. Only what is left contributes to the platform's fixed costs.
| Line item | Amount | Note |
|---|---|---|
| Order value | EUR 50 | Typical basket in the category |
| Cost of goods | EUR 27.50 | 55 percent of value |
| Packaging and shipping carried by the seller | EUR 3.75 | Partly covered by the buyer |
| Provision for returns and claims | EUR 1.50 | 3 percent of value |
| Contribution margin before the channel | EUR 17.25 | 34.5 percent of value |
| Profit the seller will not go below | EUR 5.00 | 10 percent of value |
| Theoretical ceiling for the channel | EUR 12.25 | 24.5 percent of value |
| Cost of the alternative: ads and payments in their own store | EUR 9.75 | 19.5 percent of value |
| Real ceiling | EUR 9.75 | The lower of the two |
A hypothetical example, not client data. Swap in your own numbers; the mechanics stay the same.
The theoretical ceiling here is 24.5 percent, but the real one sits five points lower, because that is what the alternative costs the seller. On your side, payment processing and dispute handling still come out of that EUR 9.75, so the contribution toward fixed costs is visibly smaller than the nominal take rate.
It pays to check what that alternative really costs, because sellers routinely understate it by ignoring maintenance and their own team's hours. The cost profile of running your own store looks different from a channel bill: high at the start and flat afterwards, instead of a percentage of every sale.
Rule of thumb: your all-in take rate has to stay below what the same order costs a seller off-platform, and the gap between those two numbers is your real advantage. Once the gap reaches zero, sellers stay out of convenience, and convenience ends at the first price increase.
How to tell your commission is set wrong
A model will not tell you that you missed by three percentage points. Seller behaviour will, with a delay, so it helps to know what to look for in the data before it reaches GMV.
Signs the rate is too high
- One-and-done pairs: a rising share of seller and buyer pairs that transact once and never return to the platform.
- Assortment filtering: your best sellers list slow-moving products with you and keep their bestsellers in cheaper channels.
- Prices higher than at source: sellers price the commission in, buyers compare and leave. At that point your commission is damaging your own competitiveness.
- Zero adoption of paid extras: when the base rate already hurts, nobody buys advertising or a higher tier on top.
- The conversation opens with a discount: new sellers negotiate the rate before they ask what the partnership includes.
Signs the rate is too low
- A price rise lands in silence: the change triggers not a single conversation and no visible churn.
- Heavy adoption of placements: sellers happily pay extra for visibility, which means they value your demand higher than you do.
- Negative contribution per order: acquiring the buyer and servicing the transaction costs more than the order leaves behind, with no path to positive.
- Cutting corners on trust: you cannot afford moderation, quality control and dispute handling, the very things that separate a platform from a classifieds page.
Why take rates rise, and how to raise yours without a revolt
Mature platforms lift their take rate almost every year, but rarely by raising the base commission. Allegro reported a Polish take rate of 12.26 percent of GMV in Q4 2025, up 0.25 percentage points year on year, with advertising alone accounting for 2.2 percent of GMV and growing faster than sales. Etsy closed 2025 at 24.2 percent and attributes revenue growth mainly to on-site ads. Fiverr lifted its marketplace take rate by 10 basis points, to 27.7 percent.
The practical conclusion across those three cases: take rates grow in layers, not in percentage points. A new layer is optional, so the seller decides whether to pay for it and can see what they are paying for. Raising the base commission hits everyone at once, including the sellers who get nothing new, and they are the ones who make the noise.
Four things to do before you raise it
- Give proper notice: the EU P2B Regulation (2019/1150) requires at least 15 days' notice of changes to terms, and longer where the change forces sellers into technical adjustments. That is a legal minimum, not a relationship standard.
- Tie the increase to new value: the higher rate arrives alongside something that did not exist before, such as payment protection, faster payouts or a new market.
- Protect your best sellers: a volume threshold or grandfathered terms for the sellers who built your liquidity costs less than losing them.
- Test on a segment: one category, one market, a clear success criterion and a decision date, rather than a change for everyone at once.
Sellers should receive their own numbers alongside the increase. Someone who measures margin rather than revenue will judge your channel the way they judge advertising, and the conversation stops being a haggle over percentage points. That shift in perspective is the subject of our piece on POAS instead of ROAS.
Defending the rate: show the bill, not the arguments
A commission is not defended with a story about platform value, it is defended with a calculation the seller can compare to the alternative. Five things are being bought:
- Demand: buyers they did not have to acquire. Usually the most expensive line and the hardest to recreate alone.
- Payments and risk: settlement, refunds, chargebacks, disputes and the cash gap between a sale and a payout.
- Buyer trust: order protection, moderation and reviews. A new seller trades on your brand before they build their own.
- Logistics and operations: if you offer them, they enter the bill with market prices as the reference point.
- Marketing: traffic from campaigns, newsletters, search and AI-generated answers.
That first line is the most expensive one for you too, because early demand has to be created by hand before any network effect kicks in. We wrote about it in our piece on the marketplace cold start problem, and it is exactly the part of the bill a seller inherits ready-made when they join a working platform.
The most effective defence is a product one rather than a sales one: a seller dashboard showing their own bill. How many new buyers they got this month, how many came back, how many impressions the platform gave them, how many disputes it handled on their behalf and what the same sales would have cost in paid media. An operator who cannot show those numbers is defending a commission with adjectives.
Disintermediation: when the transaction leaves the platform
Disintermediation is what happens when a seller and a buyer meet on your platform and settle without it. The paradox is that the better you do your job, the greater the risk. Research by Grace Gu and Feng Zhu on data from a freelance marketplace (Management Science, 2021) found that building trust between the two sides increases off-platform transactions, and that at sufficiently high trust levels this offsets the gains from better matching.
Where the risk is highest
- Repeat transactions: the same pair returns to each other every month, and after the second time the platform adds no matching value.
- High order value: on a EUR 5,000 order, a 10 percent commission is a sum worth a conversation.
- Services and long relationships: direct contact is a natural part of delivery, so it is hard to contain.
- Disappearing after checkout: if the platform plays no part once payment clears, there is nothing left to defend.
What actually works
A clause in the terms is the weakest instrument: hard to enforce and corrosive to the relationship. Economics and product work better. The cleanest example is a commission that falls with repetition: Faire charges 15 percent on repeat marketplace orders and 0 percent on orders from customers a brand brought in itself (as of 7 August 2026). The platform charges for the match, not for the mere fact of being there.
The second lever is tying value to transactions closed on the platform: buyer protection, payment terms and financing, settlement and invoicing, logistics, working history, reviews and dispute handling. The third is measurement. Minimum-value test orders, cancellations right after first contact and a drop in repeat purchases within previously active pairs all appear in the data earlier than a decline in GMV.
Practical takeaway: if leakage is growing, look first at what you are charging for. Matching is paid for once, services are paid for every month. Charging a full rate to a pair that has known each other for a year is an invitation to route around you.
A take rate is a product decision, not just a pricing one
A commission model looks like a finance decision and turns into a requirements list: multi-vendor settlement, payment splitting, payouts and invoices, rates per category and per segment, a dashboard showing value delivered, dispute and returns handling, advertising data. That is the scope we describe on our multi-vendor marketplace page. Changing the model after launch means changing settlement, which is the hardest thing to touch once other people's money flows through the platform. Growing volume adds its own pressure, which we covered in our piece on what breaks at 10x traffic.
At Seedlight we settle the take rate model when building the marketplace platform, during the Blueprint stage of the BEAM framework, together with the seller's unit economics, and only then build settlement in the Engineering phase. The order is not cosmetic: a pricing model written into the settlement architecture is cheap to change, while one bolted on after launch costs like a migration. The other side of that bill, the cost of a channel seen through a seller's eyes, is the daily work of our sister agency Amazonway in its marketplace launch service.
One honest caveat to close on: there is no single correct rate, and nobody knows yours before running the numbers for your category. What is known is the order of the questions, and the first one is about the seller, not about your revenue plan.
FAQ
What commission should a new marketplace charge?
There is no single correct rate. The starting point is the economics of one order at a typical seller: contribution margin minus the profit they will not go below, compared against what the same order would cost them off-platform. For reference, mature platforms report take rates from roughly 12 to nearly 28 percent, but those are numbers built on years of liquidity.
What is the difference between a take rate and a transaction commission?
The commission is one line in your price list; the take rate is the platform's share of total transaction value: commission plus listing fees, subscriptions, margin on payments, advertising and services. Sellers negotiate the commission, but they decide about the channel on the take rate, because that is their real cost.
How do you raise a commission without losing sellers?
Usually not by lifting the base rate but by adding an optional layer: advertising, logistics, faster payouts. If you do raise the base rate, the EU P2B Regulation (2019/1150) requires at least 15 days' notice, and longer where the change forces technical adjustments. It also pays to protect the sellers who built your liquidity and to test the change on a single segment first.
How do you reduce the risk of sellers and buyers going around the platform?
Economics beats a clause in the terms: a lower commission on repeat transactions (Faire charges 0 percent on customers a brand brings in itself) and services that cannot be recreated off-platform, such as payment protection, financing, settlement and dispute handling. Add measurement: test orders, cancellations after first contact and falling repeat purchases within active pairs.
What take rates do B2B marketplaces charge?
There are no public, comparable benchmarks for B2B: companies do not report the metric in that form and industry round-ups rarely disclose methodology. Only individual price lists are available, such as Faire, which charges 15 percent on repeat orders and 25 percent on a new customer's first order. For B2B, build the model on your own unit economics rather than on an average from an unknown source.
Journal
Co-founder of Seedlight · founder of Amazonway, marketplace sales
Newsletter
The Journal, straight to your inbox
New articles and lessons from real builds, every now and then. No spam, unsubscribe with one click.