Pricing and Take Rate: Who Pays Whom, How Much, and When to Move It
What Take Rate Is
The take rate is the fraction of every transaction that the platform keeps. If a listing sells for £100 and the platform collects £15, the take rate is 15%.
take rate = platform revenue / GMV (gross transactions)
GMV is gross merchandise value: everything that flowed through the platform. Net revenue is what the platform kept. Take rate is the ratio.
This single number tells you an enormous amount about a marketplace. Too low and the business can't sustain itself. Too high and the supply side migrates elsewhere. The healthy range depends on the vertical.
Typical Take Rates by Vertical
Rough ranges, give or take a few points per company:
Vertical Typical take rate
Ride-sharing 20 to 30%
Food delivery 20 to 30% (to restaurants), plus delivery fees
Lodging 10 to 20% (Airbnb is ~14% blended)
Secondhand goods 10 to 15%
App stores 15 to 30%
Freelance 10 to 20%
Handmade goods 5 to 15%
Auctions / used cars 5 to 10%
B2B marketplaces 2 to 10%
Stock exchanges <1%
A few observations:
- Higher-value, lower-frequency transactions (real estate, cars) have lower take rates. A 20% take on a £300,000 house would not fly
- Higher-frequency, lower-value transactions (rides, food) can sustain higher take rates because the absolute numbers are small
- Where the supply side depends entirely on the platform (Uber drivers, DoorDash couriers) the platform can take more. Where the supply side has other channels (Etsy sellers also have their own websites) the platform takes less
Who Actually Pays
Take rate is a number. The economic question is: on whom does the cost fall?
Buyer-side take
A booking fee added at checkout. The seller sees their full price; the buyer pays more than advertised. Airbnb's "service fee" is mostly buyer-side.
Pros: transparent to sellers, who tend to be more price-sensitive than buyers. Sellers set their price; platform adds its cut on top.
Cons: makes your listings appear more expensive in search compared to competitors who charge sellers.
Seller-side take
The platform deducts its cut from the seller's earnings. eBay's insertion fees and Final Value Fees are seller-side. Etsy transaction fees are seller-side.
Pros: buyer sees a cleaner price. Platform has room to extract more over time if supply is captive.
Cons: sellers feel the cost every month. They price-shop platforms if switching costs are low.
Split
Charge both sides. Uber takes from drivers (seller-side commission) and from riders (booking fee). DoorDash takes from restaurants and charges customers a service fee.
Most modern marketplaces do this because tax-wise it is cleaner to split the take across both sides, and because you can move individual levers without the other side noticing.
Ad revenue on top
Not strictly take rate, but common: charge sellers to boost their listings. eBay promoted listings, Amazon sponsored products, Etsy offsite ads. This shifts acquisition spend from the seller's own marketing budget into the platform's revenue line.
For marketplaces with weak take rates (thin margins on transactions themselves), promoted listings can be more than half the revenue. Be careful: selling ads on top of search results erodes trust. At some point users notice that every top result is paid.
Dynamic Pricing
Some marketplaces price dynamically to balance supply and demand.
Surge pricing (Uber)
When demand exceeds supply in a local area, prices rise. Two effects:
- Demand shedding: some riders opt out, reducing the queue
- Supply pull: nearby drivers head toward the surge area
Surge is a clean economic mechanism that made Uber functional. It is also unpopular with users (who see a bill that doubled during a storm) and regulators (who sometimes call it price gouging). Companies that depend on surge have to decide how much to hide from the user.
Airbnb's "Smart Pricing" does the same for nightly rates. The platform suggests a price; the host can accept or override.
Session-level personalisation
The same listing might show a different price to different users based on propensity to buy, loyalty, device type, location. Airlines have done this forever. Marketplaces walk a careful line: perceived fairness matters. Visible price differences between users are a reputation minefield.
Time-based pricing
Off-peak discounts, peak surcharges. Hotels, concert tickets, ride-sharing. Easier to explain than demand-reactive surge but less responsive.
Tiered Take Rates
Not every transaction deserves the same take rate.
- Volume discounts: sellers who do more than N transactions a month pay less per transaction. Retains your best supply
- Category differences: higher-margin categories pay more. Risky categories (high fraud, high returns) pay more. eBay does this
- Premium placement: pay a higher take rate, get preferential matching. Slippery slope; done right, a useful revenue line
- Subscriptions instead of per-transaction: Etsy Plus, eBay Store subscriptions. Predictable revenue for the platform, flat fee for high-volume sellers
Every tier is a lever. Too many tiers become confusing and game-able.
Raising Take Rate (Carefully)
As a marketplace matures and defensibility grows, most platforms raise take rate. This is legitimate: early users get subsidised to solve cold start; late users pay more because the network is valuable.
But raises are politically expensive:
- Announced raises erode trust. Sellers notice. Some leave
- Hidden raises (buried in T&C changes) erode trust more when discovered
- Gradual raises via new fee types are the most common tactic: add a "processing fee", then a "marketing fee", then a "risk assurance fee", each small, totalling a bigger effective take rate
Uber and Airbnb have both quietly moved take rates up over the years. DoorDash's effective take on restaurants has climbed past 30% in some segments, triggering regulation.
A useful rule: raise take rate in step with defensibility, not in step with need. If your platform is captive for sellers, you can raise. If they have real alternatives, they will find them.
The Race to the Bottom
In crowded marketplace categories, a competitor may undercut your take rate. Seller-side, a lower take rate is a better economic deal for them.
This can spiral. Platform A takes 20%, competitor B launches with 12%. A's best sellers migrate. A drops to 10% to retain them. B drops to 8%. Nobody makes money.
A few defences:
- Make the platform sticky. Lower take rates by themselves don't matter if switching costs are high. Review history, search rank, customer relationships built on the platform
- Focus on buyer demand. A platform that brings real buyers can charge sellers more. A platform that merely hosts listings can't
- Differentiate on non-price factors. Better tools, better payments, better fraud protection
Race-to-the-bottom dynamics are brutal. Most marketplaces that compete purely on seller-side price die.
The Free-to-Post Trap
A tempting strategy on day one: make listings free. Sellers flood in. You have supply.
The problem: free listings mean no skin in the game. Many are fake, stale, or aspirational. Buyers search, find nothing they can actually transact with, leave.
Most successful marketplaces charge some friction to list, even if not money. Identity verification, minimum listing quality, explicit commitment to fulfil. eBay's auction listings had commitment embedded. Uber drivers go through onboarding that filters unserious applicants.
Friction-free listing usually means friction-free bad listings. Add enough friction that only real sellers bother.
Price Discovery
In some marketplaces, the platform sets the price (Uber, DoorDash to an extent). In others, the seller sets it (Airbnb, Etsy, eBay listings). In others, the price is discovered through bidding (eBay auctions, live auctions, ad auctions).
The choice affects everything:
- Platform-set prices make the platform responsible for the price. Good for consistency; bad for supply-side control
- Seller-set prices push the burden to supply. Sellers experiment, find their level. But without guidance, they often get it wrong
- Discovered prices work well for unique items but are slow. They also require both sides being online at the right time or using proxies (bids)
Most marketplaces offer pricing guidance to sellers: "listings at this price sell 3x faster". This is matching-adjacent: the platform nudges supply toward liquidity.
Common Pitfalls
"We'll keep take rate low forever." You'll need to fund ops, trust and safety, matching, payments. Low take rates only work at huge volume with pure infrastructure plays. For most, 10% is the floor
"We'll raise take rate once we're dominant." You better plan to be that dominant. Until then, don't promise low rates publicly; promise to be fair
"Dynamic pricing will solve our imbalance." Dynamic pricing helps at the margin. It does not fix structural supply shortages. If you have 10 drivers and 500 riders, surge can only do so much before it looks abusive
"Buyers don't notice fees." They do, especially when checkout shows a number bigger than the product page suggested. Transparent up-front pricing beats surprise fees every time
"Our take rate should match Uber's." Uber is ride-sharing. You may not be. Benchmarks inform; they don't decide. Your margin has to work for your supply side's real alternatives
Next Steps
Continue to 06-trust-and-reputation.md to learn the mechanisms that make strangers willing to transact.