There are many great articles written about marketplace take rates.  Another great one appeared recently from CJ Gustafson of Mostly Metrics, whose newsletter I subscribe to, and I suggest you do as well.   Here’s CJ’s “money shot” of take rates from various marketplaces:

But What About B2B Marketplaces?

So why am I writing another post on this topic? Because most of these articles are, by necessity, about C2C or B2C marketplaces, not B2B marketplaces.  In the chart above, for example, only two of the twenty-five examples are B2B marketplaces — Upwork and EZCater.  (And even for these examples, their main user base is probably small businesses.)  B2B marketplaces are typically overlooked for several reasons:

• Most of the public, well-known marketplaces are C2C or B2C.
• Marketplaces thrive where buyers and suppliers are fragmented and don’t know each other, and nothing is more fragmented than C2C.
• Many B2B marketplaces are private, obscure, and/or not sexy. (Much like those of us who work in B2B software!)

Why B2B Marketplaces Are Different

B2B marketplaces differ from C2C and B2C marketplaces in a variety of ways:

• Businesses typically have more complex requirements for buying from a new supplier than consumers do, and suppliers often have to offer credit to customers.
• Businesses may want to integrate their purchase (buyer) or sales (supplier) transactions into source-to-settle and order-to-cash systems, which consumers don’t have.
• Average order values and risks can be much higher for B2B marketplaces.

The result is that the challenges, dynamics, and take rates of a B2B marketplace are quite different than those of the consumer world, and they should be mixed with caution.

What Determines B2B Marketplace Take Rates?

I’ve written about this topic a couple of times before, here and here.  But this time I took a new approach.  I built a database of approximately 60 B2B marketplaces, including public ones and those I have worked with or diligenced.  Public examples include:

• Upwork
• Fiverr
• Xometry
• ACV Auctions
• Liquidity Services
• Ritchie Brothers
• Amadeus (GDS segment)
• Sabre (GDS segment)

(The private companies shall remain nameless for confidentiality reasons.)  The database included the marketplaces’ take rates and various attributes I thought might affect them.  Then I did what any good analyst would do: I asked Claude to identify the attributes that drove differences in take rates, and I added a little judgment to the mix.  Here’s what Claude and I came up with:

Dimension	Low Take Rate	High Take Rate
Matching need and difficulty	Easy/Standardized	Hard/Complex
Process Depth	Discovery only	Full S2S/O2C
Principal Role	Agent/Facilitator	MoR/Prime Contractor
Payment Capture	Off-platform	On-platform
Average Order Value	High	Low
Fragmentation	Low on both sides	High on both sides

Let’s cover each dimension in turn:

Matching Difficulty

The harder it is to describe, find, or compare a good or service, the higher the take rate will be because the marketplace is adding more value.  This usually means that services marketplaces (e.g., complex labor), those for custom-made parts, or where the content necessary for search does not readily exist, typically have higher take rates. Xometry, a marketplace for custom-manufactured parts, for instance, has a 33.5% take rate.  Used industrial equipment sites like Liqudity Services and Ritchie Bros have take rates in the 15-25% range.

S2S/O2C Process Depth

Some marketplaces serve only as discovery tools or directories where buyers can look up suppliers, their capabilities, and inventory, but the rest of the transaction happens offline.  Most of these marketplaces cannot tell you what their take rate is because they have no idea if transactions were consummated!  These marketplaces sell eyeballs (advertising) or leads to suppliers, not transactions (outcomes).  Examples include software marketplaces like G2, Capterra, GetApp, and Software Advice (which are now all part of one company), aerospace parts marketplaces ILSMart and Partsbase, load boards like DAT, and others.  When I estimate the revenue of these marketplaces relative to buyers’ imputed purchasing intent, I typically get a “take rate” of 1% or less, but that is just a guess.

Principal Role

Some marketplaces act as facilitators or agents, while others serve as the suppliers themselves.  In the latter cases, the marketplace assumes liability, may provide a guarantee, sets the price, and becomes the vendor that the buyer adds to their vendor master.  In extreme cases, the buyer may not even know who the underlying supplier is.  Examples include the managed service marketplaces in facilities management and Xometry in customized parts.  At the other extreme, many e-procurement vendors offer marketplace-like features or offerings, but they facilitate transactions between a company and its existing and potential suppliers; they do not act as the seller of record or provide any guarantees.

Payment Capture

Most marketplaces now handle payments on their platforms, as many companies have sprung up to help them do so (e.g., Stripe, PayPal, Balance, and more), but some still do not. Adding payments increases the take rate not only directly, but also by allowing the marketplace to offer extended credit and payment terms.

Average Order Value

The higher the average order value, the lower the take rate and vice versa.  This fee structure is built into the pricing schedule of most marketplaces.  When take rates do not decline with order value, sellers get sticker shock.  ACV Auctions, for instance, used to publish a fee schedule for buyer premiums that started at 20%+ for small transactions (under $2k) and fell to 3% or less for transactions of $40k or more.

Fragmentation

I suppose this dimension could be folded into the first dimension, as it is correlated with matching difficulty.  If both sides of the marketplace are fragmented, search and discovery are more important, and neither side has bargaining power with the marketplace.  If either side is concentrated, especially the supplier side, the take rates are likely to decline. And if both sides are concentrated, you probably don’t have a viable marketplace to begin with.

A Caveat

I wrote about take rates because that is often what I am asked about.  But, as you know,  a bigger take rate is not necessarily better. You cannot put percentages in the bank, only dollars.  It’s the take rate multiplied by GMV  that produces absolute dollars of revenue, and that is what matters.  A 50 bps take rate on a $100 billion throughput is $500 million, which probably describes some of these marketplaces.  ACV Auctions gets to the same revenue with a 3-6% take rate (depending on whether you include logistics and other services) on $10 billion in volume.  Xometry acts as the seller of record, so its revenue is $500 million+, but its take is its gross margin on this volume, which is about 35%.  The key is just to avoid low take rates in low-GMV markets!