fruit stall, fruit, the old man, marketplace, farmers, night, marketplace, marketplace, marketplace, marketplace, marketplace. Twelve marketplace models examples sorted by where liquidity actually fails
Photo by EMMMMMiller on Pixabay

Industry

Part of Liquidity first: the order marketplace model problems actually arrive in

Twelve marketplace models examples sorted by where liquidity actually fails

Twelve marketplace models examples showing where liquidity actually fails, who should pay the fee in each, and the number to compute before building anything.

Marketplaces fail in a handful of recognizable ways. The shape of the thing being traded decides which one gets you. Frequency, urgency, value, and whether both parties want to meet again are the variables that matter.

Twelve marketplace models below, each one a named company. For each: where liquidity breaks, who should carry the fee, and the number to compute first.

What to take away

  • Where the same two parties will transact repeatedly, they will eventually go around you. Price and design for that from the start rather than treating it as betrayal.
  • Frequency decides whether you can build habit. Low frequency means you are buying every transaction.
  • Liquidity is a property of a segment. A market that works in one city and one category is not a working marketplace.
  • Name the segment before you write code. If you cannot list the suppliers in it by hand, it is too wide to reach liquidity.
  • Concentration bites too. At Sotheby's, the top three lots can be 60% of a year's value.

When frequency is the problem

1. Boat Trader (boats.com, YachtWorld). Boats Group runs Boat Trader, boats.com and YachtWorld. Boat Trader lists boats from owners and dealers; YachtWorld is the brokerage side, where listings often run into six figures.

Frequency Drives Fee Model

Boat Trader

Frequency
Once every several years
Habit forms
No
Fee basis
High rate on high-value item
Main risk
Acquisition cost
Key metric
Acquisition cost per transaction

Instacart

Frequency
Weekly
Habit forms
Yes
Fee basis
Low fee, repeat rate
Main risk
Disintermediation
Key metric
Repeat rate

TaskRabbit

Frequency
Once, urgent
Habit forms
No
Fee basis
Time to first response
Main risk
Slow response
Key metric
Share of requests with no response

A boat changes hands once every several years, so no habit forms and every transaction is bought again through acquisition. The fee has to carry the full acquisition cost of one transaction, which usually means a high rate on a high-value item. Compute acquisition cost against a single transaction before anything else.

2. Instacart. Instacart delivers groceries from chains including Kroger, Costco and Aldi, and its Instacart+ membership runs about $99 a year.

Groceries are needed weekly, habit forms, repeat rate carries the economics, and the fee can be low. The risk moves to disintermediation, because frequent counterparties get to know each other.

3. TaskRabbit. TaskRabbit is owned by Ingka Group, the parent of IKEA. Taskers set their own hourly rates, commonly $25 to $75 depending on the market and the skill.

Tasks arrive once, urgently, at an unpredictable moment. Supply has to be standing by, which is expensive, and requesters will not tolerate a slow response. Time to first response is the whole product here. The number to compute is the share of requests with no response inside your window.

When the two sides want each other permanently

4. Upwork. Upwork charges freelancers a flat 10% service fee, and Connects, which freelancers spend to bid on jobs, cost about $0.15 each.

Repeat Parties Change Fee Logic

Upwork

Relationship
Long-term after match
Fee
10% freelancer + Connects
Leakage risk
High
Fee should sit on
Introduction or ongoing need

Fiverr

Relationship
Trust builds over time
Fee
20% seller + 5.5% buyer
Leakage risk
Low
Fee should sit on
Platform trust and payment

eBay

Relationship
One-off strangers
Fee
13.6% final value + $0.30
Leakage risk
None
Fee should sit on
Transaction

A client and a freelancer meet once through the platform and work together for years. Charging per transaction means your revenue disappears exactly when the match succeeds. Charge for the introduction, or for something they keep needing, and say which.

5. Fiverr. Fiverr takes 20% from sellers and charges buyers a service fee of about 5.5%.

Trust takes time to build, so staying on the platform has real value because you carry the payment and the dispute record. That is a fee people will pay, and it survives the parties knowing each other.

6. eBay. eBay charges most sellers a final value fee of about 13.6% of the sale price on the first $7,500, plus $0.30 an order.

A one-off transaction between strangers: nobody is going around you, because there is no second time. The fee can sit on the transaction, and the constraint is trust rather than leakage.

When one side is scarce

7. Thumbtack. Thumbtack charges pros for each customer contact, against a weekly budget the pro sets for the categories they work in.

Scarce Side Pays

Thumbtack

Scarce side
Demand plentiful, suppliers scarce
Who pays
Pros per contact
Key metric
Active suppliers per segment
Risk
Suppliers reject rules

Gartner

Scarce side
Demand scarce, supply plentiful
Who pays
Suppliers for access
Key metric
Fill rate
Risk
Placement conflicts with best match

Nextdoor

Scarce side
Both sides thin
Who pays
Not a marketplace yet
Key metric
Segment served properly
Risk
Cold start

Demand is plentiful and suppliers have alternatives, so they will not accept your rules. The fee sits on the demand side, and the platform's job is to be the easiest route to a supplier's time. Compute active suppliers per segment, because the aggregate count hides the segments with two.

8. Gartner. Gartner sells syndicated research subscriptions to enterprise buyers and runs the Symposium conferences. A small seat count commonly runs into the tens of thousands of dollars a year.

Scarce demand, plentiful supply, so suppliers pay for access. Be careful: charging for placement starts to conflict with showing requesters the best match, and that conflict compounds. The number to watch is fill rate, which is what suffers first.

9. Nextdoor. Nextdoor verifies each user's address and shows posts to nearby neighbors. It carries a For Sale & Free section where neighbors list items at no charge.

Both sides are thin, and this is not a marketplace yet. Serving one segment properly beats being present in twenty, and the cold start problem in full is set out in marketplace models.

When the item itself is awkward

10. Sotheby's. Sotheby's is owned by Patrick Drahi's BidFair and charges buyers a premium of roughly 20% to 26% on top of the hammer price, with seller's commission negotiated case by case.

A handful of transactions carries the year, so concentration is extreme. Ten lots exist at these percentages of value:

  • 3030
  • 1818
  • 1212
  • 1010
  • 88
  • 77
  • 55
  • 44
  • 33
  • 33

The top three lots are 60%.

Losing one is a bad year, and averages are useless at this scale.

Which family of money-triggering arrangements this belongs to is worth settling first, and the shapes are set out in revenue models.

11. Chrono24. Chrono24 holds the buyer's money in escrow through Trusted Checkout until delivery is confirmed, and charges sellers a listing fee plus a commission that commonly sits in the mid single digits.

Watches need authentication before anyone will buy, so your cost per listing is real. A pure transaction fee under-recovers on listings that do not sell. Either charge for the listing or accept that the fee has to cover the failures too.

12. BizBuySell. BizBuySell sells listing packages, and its quarterly Insight Report has put median asking prices for small businesses near $300,000.

Asking prices are negotiated rather than posted, so your realized fee will differ from the posted one. Compute what you actually keep. Where competing business buyers end up on materially different terms, the price discrimination statute is worth reading with a lawyer against your own tier structure.

Reading these onto your own market

Answer three questions before writing any code.

  • How often does a given requester need this?
  • Will the same two parties want each other again?
  • Is liquidity plausible in one narrow segment you can name?

The third decides whether to start. Marketplaces that launch broadly are thin everywhere and die of it.

The economics of matching and search on electronic markets is treated in MIT's course on economics and e-commerce, and how the fee itself should be structured belongs in pricing architecture.

What the answers imply for the cost side is in unit economics.

Common questions

Can a marketplace stop parties going around it?

Contractually, sometimes, and rarely with much success. The durable answer is to keep providing something they still want after they know each other, such as payment, dispute handling, or scheduling.

Is a high fee always a disintermediation risk?

Only relative to what staying is worth. A high fee on a service people value is stable, and a low fee on a service nobody values still gets bypassed.

How narrow should the first segment be?

Narrow enough that you can name the suppliers. If you cannot list them, the segment is still too wide to reach liquidity by hand.

Which of these twelve is hardest?

Number nine, and it is also the most common starting position. Everything else on this list is a design problem. Thin on both sides is an existence problem.

More in Industry

Latest from Field Desk