The take rate is the percentage of transaction value a marketplace keeps, and it's the single price that funds the whole platform. Set it above roughly 15-20% without matching value, and sellers route around you; set it below what operations cost, and you can't reinvest in the trust and liquidity that make the marketplace worth using at all.
Quick Answer: Most healthy marketplaces charge a take rate between 5% and 30% of gross transaction value, with the right number set by how much value-added infrastructure (payments, trust, discovery, logistics) the platform provides beyond simply introducing two parties. Thin value plus a high rate is the classic setup for disintermediation.
What Determines the "Right" Take Rate
The right take rate is a function of switching cost, not intuition — it's the ceiling set by how much pain a buyer and seller would absorb to transact directly instead of through you. A rate below that ceiling captures value without triggering flight; a rate above it invites exactly the workaround it's trying to price against.
Bill Gurley's often-cited marketplace framework frames this as a simple test: if your take rate exceeds the cost of the friction you remove, you've priced yourself out of your own value proposition. A platform that just lists sellers and takes a cut, with no payments, no trust layer, and no recurring discovery advantage, is vulnerable at almost any rate above single digits.
Three variables set the practical ceiling:
- Repeat-transaction likelihood. A one-time introduction (a home renovation contractor) tolerates a higher one-off take rate than a recurring relationship (a weekly grocery delivery) that both sides would rather move offline.
- Trust and verification cost. If the platform absorbs fraud risk, identity verification, or payment guarantees, that's real infrastructure a direct deal can't replicate cheaply.
- Discovery value after the first match. If the platform keeps generating new demand for the seller (repeat search traffic), the rate is paying for an ongoing service, not a single-use referral fee.
This is the same logic behind choosing your value metric: price the thing correlated with value delivered, not the thing that's easiest to bill. For a marketplace, that thing is usually gross merchandise value (GMV) at the moment of a completed transaction — but only when completion genuinely required the platform.
Take-Rate Benchmarks by Category
Benchmarks vary sharply by category because they track how replaceable the platform's role is, not how big the company is. A capital-intensive logistics layer supports a materially higher rate than a pure listings board.
| Category | Typical take rate | Why |
|---|---|---|
| Food delivery (DoorDash, Uber Eats) | 15-30% | Owns logistics, dispatch, and last-mile risk — hard to replicate |
| Freelance/services (Upwork, Fiverr) | 5-20%, often tiered down with volume | Discovery + payment escrow + dispute resolution |
| Ride-hailing (Uber, Lyft) | 20-25% | Real-time matching, dynamic pricing, safety/insurance layer |
| E-commerce marketplaces (Etsy, Amazon third-party) | 5-15% (plus optional ad/listing fees) | Payments + discovery traffic; sellers could plausibly self-host |
| B2B/wholesale marketplaces (Faire) | 15-25% on first order, lower on repeat | High trust-building cost on first transaction, less on repeat |
| Payments-only rails (Stripe Connect-style platforms) | 0.5-3% | Pure infrastructure; no discovery or trust curation |
| Real estate/high-ticket referral | 1-3% (percentage of large transaction) | Absolute dollar value is high even at a low percentage |
The pattern: rates compress as ticket size rises and expand as the platform's operational lift rises. A $2 coffee delivery justifies a 30% cut because the alternative (walk to the store) is genuinely costly to the buyer; a $400,000 home sale can't sustain a 30% cut on pure introduction value, however much trust is involved.
Where First-Time Founders Miscalibrate
Most early-stage marketplace founders anchor to a "famous" comparable — usually Uber's ~25% — without checking whether their category shares Uber's underlying cost structure (dispatch, insurance, dynamic supply-demand matching in real time). Copying the number without copying the cost base is how the rate ends up unjustified.
Disintermediation: The Risk of Charging for Thin Value
Disintermediation happens when the take rate exceeds the ongoing value the platform provides after the first match, so both sides quietly move the relationship off-platform. It's the dominant failure mode for services and B2B marketplaces where the buyer and seller can trivially exchange contact details.
The tell is repeat-transaction leakage: first transactions happen on-platform, but the second, third, and subsequent transactions between the same two parties start happening directly. If your cohort data shows completed-transaction rate declining sharply for repeat buyer-seller pairs relative to first-time pairs, that's disintermediation, not just churn.
Three structural defenses actually work, rather than just discouraging it after the fact:
- Bundle something that can't be replicated off-platform — escrow, insurance, dispute resolution, or verified reviews that only carry weight because they're tied to the platform's identity system.
- Time-box or discount the rate for repeat transactions between the same parties, effectively admitting the marginal value of the introduction has been captured and pricing accordingly (Upwork and Fiverr both do this).
- Make the platform the cheaper path even for repeat use — payments, invoicing, and tax handling that a freelancer doesn't want to rebuild themselves, so staying is a convenience decision, not just a loyalty one.
Contractual anti-circumvention clauses can slow disintermediation but rarely stop it at scale — enforcement cost usually exceeds the recovered revenue for all but the largest transactions. Structural value beats legal deterrence.
Splitting the Fee Across Both Sides
Fee allocation between buyer and seller should follow whichever side is more price-sensitive and elastic, not an even split for its own sake — the side that reacts less to the fee should absorb more of it. This is standard two-sided-market economics (the foundational work here is Jean-Charles Rochet and Jean Tirole's research on platform pricing, which showed that optimal allocation is rarely 50/50 and depends on each side's relative elasticity and the cross-side network effect each side generates for the other).
Three common allocation patterns, each solving a different liquidity problem:
| Pattern | Example | When it fits |
|---|---|---|
| Seller-only fee, buyer sees "free" price | Etsy, Airbnb hosts historically | Buyer-side growth is the bottleneck; sellers already committed to the channel |
| Split fee, both sides pay something | Airbnb's current model (host + guest fee) | Mature marketplace balancing revenue against fee-fatigue on either side |
| Buyer-only fee (subscription or per-use) | B2B sourcing platforms, some staffing marketplaces | Sellers are scarce/high-value and would churn from any visible fee |
The side generating the larger network effect for the other side should generally be subsidized more. Early Airbnb needed hosts more than guests, so host-side friction was minimized while guest-side fees carried more of the load. As supply matures and becomes commoditized, that subsidy often flips.
Value-Added Services That Justify a Higher Rate
A higher take rate is only defensible when it's visibly funding something the seller would otherwise have to buy or build themselves — payments infrastructure, fraud protection, marketing reach, or logistics. Absent that visible trade, sellers correctly perceive the rate as pure rent extraction.
Concrete levers that move a marketplace from "thin toll" to "worth the cut":
- Payments and instant payout — absorbing PCI compliance, chargebacks, and international payout complexity is real infrastructure cost most sellers can't replicate solo.
- Verified trust signals — reviews, identity verification, and background checks that buyers demonstrably weight in conversion, not decorative badges.
- Demand generation beyond the seller's own reach — paid acquisition, SEO-driven organic search traffic, and repeat-visit habits that keep bringing new buyers to an existing listing.
- Optional paid upgrades layered on top of the base take rate — promoted listings, analytics, or premium support that sellers can choose into rather than being forced to absorb.
- Dispute resolution and guarantees — a buyer-protection guarantee that increases conversion rate enough to offset the fee, provided it's actually invoked and honored, not just marketed.
This is the marketplace version of value-based vs. cost-plus pricing: the take rate should track the bundle of services delivered, not simply "whatever percentage covers our costs plus margin." Sellers will tolerate a materially higher rate when they can point to a specific service it buys.
The Network-Effect Loop a Wrong Rate Can Break
A take-rate increase doesn't just cost you revenue at the margin — it can break the network-effect loop that made the marketplace valuable in the first place, because supply and demand reinforce each other and a rate hike that pushes out even a modest slice of supply thins the loop for every buyer left behind.
The loop, simplified: more sellers → better selection and lower prices → more buyers → more transaction volume → more sellers attracted by that volume. A take-rate increase that pushes marginal sellers to leave (or raise prices to compensate) thins selection, which softens buyer growth, which reduces the volume that justified the original seller base — a reinforcing loop running in reverse.
Watch specifically for these early warning signals before a rate change compounds:
- Seller churn concentrated among your best-reviewed or highest-volume sellers first — they have the most outside options and leave first.
- Rising customer acquisition cost per buyer as selection thins and organic word-of-mouth slows.
- A widening gap between listed price and effective price as sellers quietly pass the fee increase to buyers, which dulls the marketplace's price-competitiveness story.
Key Takeaways
- Take rate should track ceiling of replicable friction, not an arbitrary industry-standard percentage — price against what going direct would actually cost the two sides.
- Benchmarks range from under 1% (pure payment rails) to 25-30% (full-logistics marketplaces like food delivery), with the driver being operational lift, not company size.
- Disintermediation risk rises sharply when the platform's value is front-loaded into the first match and thin afterward — repeat-transaction leakage in your data is the earliest reliable signal.
- Fee splits should follow relative elasticity and cross-side network effects, not a default 50/50 — subsidize the side generating more value for the other.
- A higher rate is only defensible when tied to a visible, named service — payments, trust, demand generation, or dispute resolution — that sellers would otherwise pay for elsewhere.
- A take-rate change ripples through a feedback loop across two populations with a time lag, so model it as a system before shipping it as a line-item change.
Frequently Asked Questions
What is a good take rate for a marketplace startup?
Most early-stage marketplaces start in the 5-15% range and adjust once they can prove a specific value-added service justifies more. Starting low protects supply-side liquidity while the marketplace is still building trust and repeat-usage habits, which matters more early than maximizing per-transaction revenue.
How do you calculate a marketplace take rate?
Take rate is calculated as platform revenue divided by gross merchandise value (GMV) for a given period, expressed as a percentage. If $100,000 in GMV generates $12,000 in platform revenue across commissions and fees, the effective take rate is 12%, even if the headline commission is a flat number layered with add-on fees.
Why do marketplaces lower take rates as they scale?
Take rates often compress at scale because larger sellers have more outside options and more negotiating leverage, and because a marketplace's marginal cost of serving an additional high-volume seller is lower per transaction. Tiered or volume-discounted rates (common on Upwork and Fiverr) keep high-value supply from churning to competitors or going direct.
Can a marketplace charge different take rates to buyers and sellers?
Yes, and most mature marketplaces do, splitting the fee based on which side is more price-sensitive and which side generates more value for the other. This is standard two-sided-market pricing theory, not an inconsistency — the goal is maximizing total transactions, not charging each side identically.
Is a 0% take rate ever the right strategy?
A 0% or near-0% take rate can make sense temporarily to solve a cold-start liquidity problem, monetizing later through value-added services (ads, premium listings, payments) once volume exists. This mirrors the sequencing logic in freemium vs. free-trial thinking: give away the core match, monetize the layer built on top of proven usage.
For the broader pricing playbook this fits into — value metrics, packaging, and monetization sequencing beyond marketplaces — see the pricing and monetization complete guide. And because a take-rate decision is ultimately a bet on how sellers and buyers behave, it's worth grounding in the underlying jobs to be done each side is hiring the marketplace for, and mapping where in the customer journey the fee actually gets felt.