What Is Take Rate in a Marketplace?

By Brian Tr · Updated Oct 6, 2026

Two marketplaces can each handle a billion dollars of transactions in a year and report wildly different revenue. One keeps a few cents on the dollar. The other keeps more than a quarter. The ratio between what flows through and what the company keeps is the take rate, and it's the single number that most often gets quoted, compared and misused when people talk about marketplace economics.

This article is the hub for the monetization section of our marketplace series. It defines take rate precisely, shows the variants in the numerator and denominator, works through a calculation, and explains what pushes a rate up or holds it down. It then lays out what several public marketplaces report in their own filings, with each company's definition noted, because the definitions are not the same. We keep the short version of GMV versus revenue in What Is a Marketplace Business Model?, so we won't repeat it here. The goal is the deeper layer.

The definition, and why it's slippery

In its plainest form, take rate is revenue divided by the value of the transactions that produced it:

Take rate = marketplace revenue / gross transaction value

Bill Gurley, in his 2013 essay on marketplace pricing, called the same idea the rake. He defined it as the percentage of gross merchandise sales the marketplace charges, and used eBay's 2011 numbers as the example: roughly $6.6 billion of revenue against roughly $68.6 billion of sales, a rake just under 10%.

That looks simple. The trouble is that both halves of the fraction can be defined several ways, and companies pick the version that suits their business.

Numerator variants

• Commission only. Just the fee taken on each completed sale. This is the cleanest measure of what the core transaction earns.
• Marketplace revenue. Commission plus other fees tied to the transaction itself, such as buyer fees, payment fees or listing fees.
• Total revenue. Everything, including advertising, subscriptions, shipping labels, financing and any non-marketplace lines. Etsy's take rate, for example, is revenue divided by GMS, and revenue there includes services such as ads and payments, so it's a total-revenue rate.

Denominator variants

• GMV or GMS. The total dollar value transacted. Some companies include taxes, tips, shipping and fees in it, and some don't. Airbnb's Gross Booking Value, for instance, includes host earnings, service fees, cleaning fees and taxes, net of cancellations. DoorDash's Marketplace GOV includes taxes, tips and consumer fees.
• Timing. GMV is usually counted when the order is placed. Revenue may be recognized later. Airbnb's filing says revenue is recognized when guests check in, so a booking made in December for a stay in March sits in one year's GBV and another year's revenue.

Gross versus net revenue

There's one more layer, and it changes the headline number completely: accounting treatment. If a company is the principal in a sale, it controls the good or service and records the full price as revenue. If it's an agent, it arranges the sale and records only its fee. Uber's 10-K says it has concluded it's an agent in its driver and merchant arrangements, so revenue is net of driver and merchant earnings and driver incentives. A marketplace that buys inventory and resells it can report a revenue figure that's a large share of GMV, but that's a reseller-style margin, not a take rate in the marketplace sense.

The practical rule: before comparing two take rates, check what's in the top, what's in the bottom and whether revenue is net or gross. A computed ratio that mixes definitions is not a benchmark.

How to calculate it: a worked example

Here's an illustrative marketplace. The numbers are made up to show the arithmetic and aren't from any company.

Suppose a services marketplace processes $10,000,000 of gross transaction value in a year. It charges sellers a 10% commission, sells $300,000 of promoted listings, and earns $200,000 from optional seller tools and payment fees.

Three different "take rates" come out of the same business:

• Commission take rate: $1,000,000 / $10,000,000 = 10.0%
• Marketplace-plus-services take rate: $1,500,000 / $10,000,000 = 15.0%
• Take rate if the company booked the whole sale as principal: revenue would be about $10,300,000 and the "rate" would look enormous, but the company would also carry about $8,800,000 of cost of supply. Gross profit would still be about $1,500,000, so the economics are unchanged and only the label moved.

The third case is why serious comparisons use net revenue or gross profit as a share of GMV. For how revenue turns into margin, see our gross margin reference.

One more point before leaving the arithmetic: take rate is a ratio of flows, not a profit measure. A 20% take rate that's consumed by 25% of GMV in payment costs, incentives and support is a worse business than a 10% rate with lean operations. Take rate tells you what the marketplace captures. It says nothing about what it costs to capture it.

What determines how high a take rate can go

There's no fixed number a marketplace is "supposed" to charge. The ceiling comes from what buyers and sellers are willing to pay for the intermediary, compared with what they could do without it. The factors below are our synthesis of Gurley's argument plus standard marketplace reasoning, and the rest of this section labels which is which.

1. Value added per transaction. Gurley's central test is that a sustainable marketplace is one where the value of being in the network clearly outweighs the transactional cost. If the marketplace finds the customer, vets the supplier, guarantees the outcome and handles payment, it has more to charge for than one that only lists. This is the main driver.

2. Managed versus unmanaged. A managed marketplace, where the operator curates supply, guarantees quality or handles fulfillment, usually supports a higher rate than an open listing board, because it does more work and takes more risk. That's reasoning, not a rule: a managed model also has higher costs, so a higher rate doesn't mean a higher margin.

3. Frequency of the transaction. Fees paid on a one-off purchase are less visible than fees on a weekly recurring one. When the same buyer and seller transact repeatedly, the fee adds up and the incentive to take the relationship off the platform grows. Marketplaces with repeat relationships tend to need ongoing value, such as payment protection and scheduling, to justify what they charge.

4. Ticket size. A 15% fee on a $20 order is $3. On a $20,000 project it's $3,000. Large tickets make the fee more noticeable and make workarounds more attractive. Our note on B2B versus B2C marketplaces covers how deal size shapes marketplace design.

5. Competition and multi-tenanting. If sellers list on several marketplaces and buyers compare across them, a high rate on one is easy to avoid. When a marketplace has strong network effects and its buyers are hard to reach elsewhere, it can charge more. Where users can multi-home cheaply, the rate tends to be pushed toward the lowest competitor's.

6. Disintermediation risk. Gurley's essay argues that a high rake becomes part of the buyer's final price and gives suppliers a reason to look elsewhere, which can threaten the marketplace's own sustainability. His examples include a photo marketplace paying content owners only 30% of receipts, which he called potentially fragile, and a daily-deals business whose rake of about 38% (after merchants underwrote a 50% discount) left little room. Note that those are 2013 observations by one investor, not current figures.

7. The alternative price setter. Gurley also tells the story of Booking.com, which he says started with a 10% agency-model commission and used the lower rake to win suppliers in Europe. He adds that a disruptor can attack a high-rake leader by pricing lower, and cites Taobao against eBay in China as an example. The lesson in his closing argument is restraint: charging what the marketplace can extract is riskier than charging what it should.

The seven factors interact. A high-frequency, high-ticket, easily multi-homed marketplace is the hardest place to hold a high rate. A low-frequency, managed, trust-heavy one is the easiest.

Take rates in public filings

Public companies give us real numbers, and also real examples of how different the definitions are. The table below uses each company's own latest annual figures that we could verify, all for fiscal 2025. Where we computed the ratio ourselves from reported inputs, we say so.

Sources: Etsy Q4 and full-year 2025 release, Fiverr FY2025 20-F, Upwork FY2025 10-K, DoorDash FY2025 10-K, Uber FY2025 10-K, Airbnb FY2025 10-K, eBay Q4 and full-year 2025 results, Grab FY2025 20-F.

A few notes keep the table honest.

• The computed rows are not the companies' own metrics. Uber and Airbnb don't publish a "take rate" in the filings we read. Our ratios divide total revenue by the volume metric each company reports, and Uber's Gross Bookings includes Freight. Treat them as rough indicators.
• Definitions differ in what's counted. Uber's Gross Bookings is defined as the total dollar value of rides, delivery orders and Freight revenue, including taxes, tolls and fees, before consumer discounts and driver incentives. Grab's incentives, which its filing puts at about 10.2% of on-demand GMV in 2025, reduce its revenue. A marketplace that spends heavily on incentives will show a lower net rate than its fee schedule suggests.
• Rates reflect the business, not just the pricing. Services and goods marketplaces such as Fiverr and Etsy are in the mid-20s, ride-hailing in Uber's case is near 27% on this measure, and delivery and travel sit near 13 to 14%. Part of that is accounting (net versus gross, what's in the denominator) and part is what each marketplace does for the transaction. We wouldn't read the table as a ranking of how greedy each company is, and we wouldn't set a target rate from it.

How take rates change over time

Take rates are rarely static. The filings above show three patterns.

Rates rise as the marketplace adds paid services. Etsy's take rate went from 22.3% to 24.2% between 2024 and 2025, and its release attributes the revenue growth mainly to on-site advertising. In the same year its GMS fell 5.3%, so revenue grew while the volume underneath it shrank. Upwork's marketplace take rate rose from 15.4% in 2023 to 18.7% in 2025, and its 10-K credits product initiatives for the expansion. Rising take rate on flat volume is how monetization maturity looks in a filing, and it's also a reminder that revenue growth can come from charging more on the same activity.

Rates move when the fee structure changes. Airbnb's filing says that in October 2025 it began moving from a split-fee structure, where both host and guest pay a service fee, to a single fee charged only to the host. Fiverr's 10-K attributes a small rise in its marketplace take rate to "slight adjustments in fee structure." The headline rate can shift without anyone's behavior changing, because the fee has moved from one side to the other or been restructured.

Rates are defended carefully. Fiverr's risk factors warn that changes to its take rate could reduce marketplace activity, and that it may be forced to lower the rate if the platform's perceived value falls. That's the same tension Gurley described: the best rate is the one users will keep paying.

The four main ways marketplaces raise revenue per transaction without touching the headline commission are covered in their own articles, so we only summarize them here.

• Revenue-model choice. Commission, subscription, listing fee, lead fee and buyer-side fee each produce a different take rate and different incentives. See marketplace revenue models.
• Payments. Handling the money creates a fee line, keeps the transaction on the platform and gives control over payout timing. See marketplace payments.
• Value-added services. Insurance, financing, shipping, tools and verification add revenue and give users a reason to stay after the match. See value-added services for marketplaces.
• Advertising. Paid placement monetizes crowded supply without raising the commission. See marketplace advertising revenue.

The common sequence is to start low to win supply, build density and liquidity, and then layer services and ads on once sellers are competing for attention. Raising the headline commission is the riskiest of these because every seller sees it at once.

Practical guidance for setting a take rate

These are rules of thumb that follow from the reasoning above, not data.

1. Define it before you report it. Write down the numerator, denominator and treatment of taxes, shipping, refunds and incentives, and use the same definition every period.
2. Track more than one version. A commission rate and a total-revenue rate tell different stories. A rising total rate on a flat commission means services and ads are doing the work.
3. Net out the costs. Look at contribution after payment fees, incentives and support, not only the headline rate.
4. Segment it. A blended rate hides categories, markets and seller tiers that sit far above or below the average.
5. Price to the value, then test the ceiling gradually. Gurley's logic implies a rate that leaves buyers and sellers better off than their alternatives. Changes are easier to absorb when they're tied to new value than when they're a straight increase.
6. Watch seller behavior, not just revenue. Listings moving to other channels, falling seller activity and off-platform deals are early signs the rate is too high for the value delivered.
7. Mind the sequence. In a new market, supply and demand are the constraint, and the order in which you recruit them matters too. See supply-first versus demand-first.

Key Facts

• Take rate is revenue divided by gross transaction value. Bill Gurley's 2013 essay defines it as the percentage of GMS the marketplace charges, citing eBay's 2011 rake as just under 10%.
• Definitions vary. Etsy defines take rate as revenue divided by GMS, which includes services and ads, while Fiverr and Upwork report a narrower marketplace take rate based on marketplace revenue only.
• Etsy's FY2025 take rate was 24.2%, up from 22.3% in 2024, while GMS fell 5.3% to $11,916.9 million.
• Upwork's marketplace take rate was 18.7% in 2025, 18.0% in 2024 and 15.4% in 2023.
• DoorDash's Net Revenue Margin, revenue as a percentage of Marketplace GOV, was 13.4% in 2025 and 2024.
• Uber's 10-K states it acts as an agent, so revenue is net of driver and merchant earnings and incentives. Dividing its FY2025 revenue by Gross Bookings gives about 26.9% (our computation).
• Airbnb began moving to a single host-paid service fee in October 2025, an example of a fee-structure change that shifts the measured rate.
• Take rate measures what a marketplace captures, not what it keeps. Incentives, payment costs and support come out of it.

Related Reading

• Marketplace Revenue Models
• Marketplace Payments
• Value-Added Services for Marketplaces
• Marketplace Advertising Revenue
• What Is a Marketplace Business?
• Why Density Beats Reach in Local Marketplaces
• Managed Marketplace
• B2B vs B2C Marketplaces
• What Is a Marketplace Business Model?