The Economics of Marketplaces: Take Rates, Middlemen, and Power

Summary

Podcast Summary: Run the Numbers

Episode: The Economics of Marketplaces: Take Rates, Middlemen, and Power

Host: CJ Gustafson
Date: March 2, 2026

---

Episode Overview

In this episode, CJ Gustafson dives deep into the economics and evolution of marketplaces—those business models that sit between buyers and sellers, take a cut (the "take rate"), and attempt to add value to both sides. CJ examines why the "cut out the middleman" tech myth is overstated, unpacks the mechanics behind successful and failed marketplaces, explores take rate dynamics, and discusses how both history and technology (especially AI) shape who “gets to be the middleman” in every era. The episode draws vivid parallels between Mesopotamian traders and modern online platforms, making the point that the coordination function of the middleman persists, even as the tools and economics change.

---

Key Discussion Points & Insights

1. The Middleman Myth: Why It Endures

• Middleman as Power Broker:

  • Modern tech companies like Amazon, Airbnb, and eBay don't make or own the inventory—they exist to facilitate transactions and extract a cut.
  • "Everybody thinks technology eliminates middlemen. That’s the myth." (A, 00:00)
  • The “middleman” is the solution to a coordination problem, not the problem itself.

• Platform Power Is About Controlling Transactions:

  • Empires have been built on tiny take rates; sometimes, these same cuts bleed the systems that support them.

---

2. Historical Evolution of Marketplaces

• From Ancient Bazaars to Amazon:
  • Early marketplaces provided structure and trust for commerce, a function echoed in today’s platforms.
  • "The concept evolved into the Persian Bazaar, a linear strip of stalls… I'm a big strip mall guy myself." (A, 01:36)
• Categorization as an Innovation:
  • The Greek Agora grouped vendors by type, a forerunner of category-based digital marketplaces.
• Permanent Retail as a Milestone:
  • The Trajan markets in Rome were multi-level, permanent shopfronts—think “mall meets platform.”
  • "If an ancient Roman could have only tasted a Cinnabon, maybe Rome would have never fallen." (B, 02:43)

---

3. The Digital Revolution: How Marketplaces Modernized

• From Bulletin Boards to Online Giants

  • Pre-Amazon attempts (like Boston Computer Exchange in 1982, Bookstacks Unlimited in 1992, PizzaNet in 1994) set the stage for the online commerce boom.
  • But Amazon won with convenience and customer focus, later layering in the now-iconic 15% referral fee (A, 04:03).

• Trust as a Selling Point:

  • eBay’s revolutionary feedback system and PayPal acquisition enabled global, trusted peer-to-peer trade (A, 05:39).

• Craigslist as the "Cockroach" of the Internet:

  • Drove classifieds online but stuck to discovery only, without payments or trust layers.
  • Led to the “unbundling” of generalist marketplaces and rise of verticals (A, 06:11).

---

4. The Rise of Vertical Marketplaces

• Serving Narrower Niches, Better:

  • Zappos, StockX, Hipcamp, Reverb, and Fair carved out categories neglected or generically served by aggregators.
  • They expanded markets by professionalizing trust, curation, and financial tools, often adding adjacent value (A, 07:23 – 08:20).
  • "They turned a gray market into a transparent one, and the transparency itself actually grew the market." (A, 07:50)

• Expanding by Solving for Both Sides:

  • Hipcamp created new supply by enabling private land camping.
  • Fair de-risked retail buying by allowing free returns and flexible terms.

---

5. The Mobile & App Era: Lowering Friction, Raising Stakes

• Mobile Apps as Marketplace Accelerant:

  • The iPhone (2007) and App Store (2008) collapsed the time between intent and transaction, unlocked new supply, and cemented mindshare as the key battleground (A, 14:29).
  • "An entire generation of marketplaces didn’t just benefit from mobile. They couldn’t have existed without it." (A, 15:08)

• Mindshare Is the Real Long-term Moat:

  • Platforms win and stay won by becoming the go-to app consumers use without thinking—pushing rivals into irrelevance (A, 16:09).
  • "You aren't competing against other CFO podcasts, you're competing against Taylor Swift." (A, 16:55)

---

6. Dissecting Take Rates: The Hidden Math

• CJ’s Take Rate Formula:

  • "Take rate is a function of purchase frequency, ticket size, and platform labor intensity." (A, 18:09)
    • Higher purchase frequency = lower feasible take rate
    • Higher average ticket size = lower percentage take
    • More operational/labor intensity = higher justifiable take
  • "Nobody is paying a 20% fee on a $500,000 home sale... That’s why real estate brokerages hover around 5 to 6%." (A, 19:09)
  • Logistics-heavy operations (Uber, DoorDash) justify higher takes, trusted platforms (Etsy, StockX) sit in the middle, while pure lead gen (Thumbtack) stays low.

• Take Rate Examples and Cautionary Tales:

  • Booking.com won by dramatically undercutting Expedia/Travelocity's take rate (A, 22:49).
  • Groupon’s effective take stretched to 70% when paired with deep discounts—a non-sustainable “hostage situation” for merchants (A, 23:30).
  • "There’s a big difference between what you can extract and what you should extract. Water runs downhill." (A, 24:28, quoting Bill Gurley)

---

7. When Take Rates (and Fee Stacking) Break the Experience

• Airbnb’s Growing Pains:
  • Fee stacking (cleaning, service, processing) made the actual cost much higher than advertised—turning Airbnbs back into the same priced, less convenient hotel alternatives (A, 28:15 - 30:31).
  • "The platform that was supposed to make travel cheaper and more personal had somehow become more expensive and more annoying than a Holiday Inn." (A, 29:44)
  • Course corrections: Airbnb eventually pushed hosts to bake fees into rates and improved total price display.

---

8. The Marketplace+ Model: Beyond Transaction Cuts

• Adding Software and Services:

  • Marketplaces win durable ARPU by embedding themselves in supply-side (or demand-side) workflows:
    • OpenTable started as a SaaS seating chart before opening to consumers (A, 31:24).
    • Outdoorsy built SaaS for RV rental operators before aggregating listings.
    • PartsTech added labor estimate tools to its parts marketplace, tripling ARPU (A, 32:11).
    • Amazon Prime transitioned transaction-based business into subscription/retention engines.

• Pattern: Solve adjacent pain points before, during, or after the core transaction to increase attachment and reduce churn (A, 32:11).

---

9. The Limits of Trust Layers: Facebook Marketplace as an Outlier

• Success With Minimal Trust Infrastructure:
  • Despite nearly 8x eBay’s user base and massive gross merchandise value, Facebook Marketplace adds little in terms of payment, buyer protection, or grading (A, 33:30 – 34:56).
  • "The transaction flow... is just like, hey, meet me in this Wendy’s parking lot... hope I can come back." (A, 34:56)
  • Scam rates are high, but volume is massive—proof that trust can be bootstrapped by identity alone.

---

10. Common Causes of Marketplace Failure

1. Becoming a Retailer (Inventory Traps):

   • When platforms start holding inventory, their risk, cost base, and exposure all change.
   • "The whole beauty of a marketplace is that you facilitate the transaction without owning the thing being sold." (A, 37:04)
   • Case study: Hodinkee—grew from content and affiliate commissions to inventory, got crushed by a watch market downturn, and faced deep layoffs and fire sale exit (A, 37:22 – 39:52).

2. Disintermediation:

   • Both sides of the marketplace can find incentives to "cut out the middleman" after initial introduction (A, 40:09).
   • Homejoy (cleaning services) couldn’t prevent loss of repeat biz due to offline switching; Thumbtack survived by shifting to a lead-gen model (A, 41:20).

• Key Diagnostic:
  • "Does the platform add value to every transaction, or only the first one?" (A, 42:02)
  • Platforms that are valuable every time (Uber) win; those that add value just once become “toll booths” and fade.

---

11. The Future: Marketplaces, Cars, Homes, and AI

• Marketplace Penetration in Big Purchases:

  • Carvana proved people will buy high-ticket items sight unseen for convenience (A, 43:41).
  • Home buying is following, but slowly—inspection and in-person “gut check” are harder to displace (A, 44:36).

• AI’s Impact on Marketplaces:

  • AI will make discovery and matching personalized and instantaneous ("AI will replace browsing with matching," A, 45:17).
  • The power to sell premium placement/ads will be challenged as AIs deliver the best match directly to buyers.
  • AI tools will make starting and operating a business on marketplaces much easier, driving up supply and competition (A, 46:02).
  • AI-powered matching could solve service platform disintermediation: if the system always re-pairs you with the best match, no reason to go outside (A, 46:46).
  • "The recommendation engine is the product." (A, 46:59)

• Trust & Safety:

  • AI can automate fraud detection and make platforms safer, expanding the user base (A, 47:13).

• What AI Won’t Change:

  • The need for a middleman—someone must coordinate, guarantee trust, and settle disputes (A, 47:43).

---

12. The Unchanging Fundamentals

• The Middleman Always Returns:
  • "You can't cut out the middleman. You can only become a better one." (A, 49:06)
  • Platforms must always add more value than they extract, avoid becoming de facto retailers unless prepared for risk, and build something both sides need every time (A, 48:50).

---

Notable Quotes & Memorable Moments

• On Marketplace Collapse:

  • "Marketplaces rarely collapse in dramatic fashion. They collapse when the balance shifts, when the cut outgrows the value. And by the time that shift shows up in the numbers, it’s already too late." —A, 01:09

• On Fee Frustrations:

  • "You stared at 287 bucks. Where did the other 137 come from? A $75 cleaning fee, a $42 service fee, an $18 Airbnb processing fee, which, I’m sorry, like what? What is that?" —A, 28:29
  • "The thing you disrupted is now actually the better deal." —A, 29:14

• On Take Rate Restraint:

  • "There’s a big difference between what you can extract and what you should extract. Water runs downhill." —A, 24:28

• On Tech Cycles:

  • "Time is a flat circle and we just go from bundling and unbundling." —A, 13:41

• On Mindshare as Moat:

  • "You aren't competing against other CFO podcasts, you're competing against Taylor Swift." —A, 16:55

• On AI and Marketplace Evolution:

  • "The recommendation engine is the product." —A, 46:59

• Episode Closing:

  • "You can’t cut out the middleman. You can only become a better one." —A, 49:06

---

Timestamps for Key Segments

| Segment | Timestamps | |------------------------------------------------|-----------------| | Middleman myth, ancient history | 00:00 – 03:09 | | Digital origins, early online platforms | 03:09 – 05:39 | | Trust, reputation, verticalization | 05:39 – 08:26 | | Modern mobile era, mindshare | 13:21 – 17:53 | | Take rate math and spectrum | 18:00 – 21:14 | | Cautionary tales, when take rates go wrong | 21:50 – 30:31 | | Marketplace+ model, SaaS hooks | 31:24 – 33:29 | | Facebook Marketplace, raw scale, no trust | 33:29 – 36:14 | | Marketplaces’ fatal flaws: inventory, disintermediation | 36:23 – 42:35 | | Future: Cars, homes, AI disruption | 42:41 – 47:43 | | The eternal middleman, closing | 48:12 – 50:32 |

---

Takeaways

• Marketplaces don't eliminate the middleman; technology just changes who and how.
• Take rates must balance value and extraction, or platforms become victims of their own greed.
• Deep specialization, trust, and operational layers are the keys to outlasting generalist platforms.
• AI will accelerate existing trends, lowering supply friction and raising competitive pressure, but not upending the core platform value proposition.
• The biggest threats: getting stuck with inventory (becoming a retailer) and failing to justify recurring value (getting disintermediated).
• Mindshare—the spot on a consumer’s app home screen—is the endgame moat.
• The middleman may change, but the role is as vital as ever: "You can’t cut out the middleman. You can only become a better one."

---

Summary compiled and structured to capture all core themes and lessons, with direct pull-quotes for flavor and accuracy. Skip the episode intro/ads; jump right into the middleman debate, ancient commerce, digital platform mechanics, and the enduring logic of take rates and platform dominance.

Transcript

[A]

Everybody thinks technology eliminates middlemen. That's the myth. Some of the most powerful companies in the world don't make anything. They don't own the cars, the homes, or the inventory. They sit in the middle of the transaction and take a cut. Sometimes that cut builds an empire. Sometimes it slowly poisons the entire system.

[B]

I'm in the empire business.

[A]

Amazon started by shipping books on a concrete floor and turned a 15% referral fee into infrastructure. We are a famously unprofitable company. EBay. They made strangers trust each other and built a global bazaar out of reputational scores. I don't think ebay wants to sell it and I don't think ebay should. Airbnb unlocked spare bedrooms, but then nearly priced itself into looking like the Marriotts it wanted to displace. Then you have Facebook, who built one of the largest peer to peer marketplaces on earth, and they barely layered in machinery everyone else depends on. How about now? You're still wired in. Same core model, radically different outcomes. That's what we're getting into today. In this video, we're going to break down the mechanics underneath those stories. The hidden math behind take rates, the inventory trap that turns platforms into retailers, the disintermediation problem that quietly kills service marketplaces, and the mindshare war that is raging on your app home screen. Because marketplaces rarely collapse in dramatic fashion. They collapse when the balance shifts, when the cut outgrows the value. And by the time that shift shows up in the numbers, the it's already too late. You got me. This is a cfo explains marketplaces. Marketplaces have been around since the grain traders first met on the Mesopotamian plains. Historians trace the earliest structured markets back to 3000 BC in the cities of Ur and Uriq, located in modern day Iraq. This was the birth of the designated area. The radical idea that trade shouldn't just happen anywhere but at a specific coordinate. Like, yo, let's all meet up at this place at a specific time and do all this commerce stuff. The concept evolved into the Persian Bazaar, a linear strip of stalls covered for protection from the sun. I'm a big strip mall guy myself. There was this Olympia Sporting Goods I used to always hit at the Yarmouth Cape Cod Patriot Plaza growing up. And honestly, the DNA is exactly the same.

[B]

Strip malls are as American as hot dogs and fourth July fireworks.

[A]

The Greeks eventually added the Agora, which introduced categorization, grouping stalls by product type so you didn't have to wander past fish to find clothes. And by 100 AD, the path was paved for the Hollister and Abercrombie era with the Trajan markets in Rome, these are the earliest examples of a permanent multi level retail shop front. Can you believe that?

[B]

Yeah. I mean, if an ancient Roman could have only tasted a Cinnabon, maybe Rome would have never fallen.

[A]

All right, Ben, let's fast forward from ancient Rome to the digital shift of the 1990s. Amazon was the first real online marketplace. Well, kind of technically, if we're going to be pedantic here, the path was cleared by a few outliers you've probably forgotten. In 1982, the Boston computer Exchange launched as a dial up bulletin board for used computers.

[B]

Yeah, dial up bulletin boards. That sounds extremely satisfying. If you love busy visuals, I'm sure you'd love the earth shattering sound in your eardrums from dial up.

[A]

I hate that sound. I still remember it as a kid with aol. Do you remember when your house first got that?

[B]

Yeah, I do, I do. You couldn't make a phone call and use the Internet at the same time?

[A]

Nah, I'm trying to go on aim. In 1992, Bookstacks Unlimited was selling books via dial up before moving to the web as books.com. then on August 11, 1994, a site called Net Market conducted what is widely considered the first secure retail transaction on the World Wide web when a student bought a copy of Stings 10 Summoner's Tales for $12.48. I guess inflation didn't hit yet. Even Pizza Hut beat Amazon to the punch, launching Pizzanet in 1994 so people in Santa Cruz could order a large pepperoni pizza online.

[B]

Hello?

[A]

But Amazon was the one that actually moved the needle. It succeeded not by being the first, but by being the most convenient and customer centric. In its early days, Amazon operated as a digital middleman with almost zero inventory. When a customer ordered a book, Jeff Bezos and his team would order it from a wholesaler like Ingram Content Group, pack the box on their hands and knees on a concrete floor and drive it to the USPS themselves. Now, they didn't have a take rate yet because they were a first party retailer, buying at wholesale and selling at a slim margin. It wasn't actually until 1999 that they launched the third party marketplace place setting, the 15% referral fee that still dominates the industry.

[B]

It's crazy that their, their take rate hasn't changed that much. I mean, how long did it take them to make a profit?

[A]

And we're actually going to get to that in a little bit. Because while Amazon was Perfecting the everything store, Craigslist was becoming the cockroach of the Internet. Launched in 1995, it gutted the newspaper revenue by making classified ads free. Craigslist was the discovery layer. It didn't handle payments or shipping. It simply introduced you to a stranger. It forced the world to perfect the meat in a Starbucks parking lot with a $20 bill transaction. And if Craigslist was about discovery, eBay was the trust architect. Founded as Auction Web in 1995, eBay tackled the stranger danger problem of the early web by inventing reputation as a currency. Ebay's feedback form turned trust into a data point, proving that people would trade globally if their reputation was on the line.

[B]

And then they went a step further by acquiring PayPal.

[A]

Ooh, a student of history, I see. Yes, by acquiring PayPal in 2002, they removed the friction of mailing paper checks and made digital commerce feel like a professional exchange rather than just this gamble. However, being a generalist eventually became a weakness for ebay and also Craigslist. Because they tried to be everything to everyone, they couldn't be the best at any one thing. This led to the unbundling of Craigslist, a concept famously mapped out by investor Andrew Parker.

[B]

Ah, the Gong show guy. I'm excited to see. Maybe in like 10 years, do you think we'll see the same sort of kind of unbundling map with AI or ChatGPT or something like that?

[A]

I just want the Gong show to come back, Ben, but I think we will actually with ChatGPT and all the applications underneath you, you actually already see it with, with certain vertical specifics like legal, you have Lagora, you have Harvey, you had Clio, which came before. But you're starting to see the unbundling of vertical niche use cases within SaaS. So yeah, there is a parallel. Entrepreneurs realized that if they took just one category, like housing or jobs, and made it safer and more specialized, they could build a multibillion dollar empire. This gave rise to the vertical marketplace. At the time, many skeptics thought these opportunities were simply too small. It fell into the small market fallacy. They asked just shoes or who'd buy a used sneaker online. But players like Zappos and Stockx proved them wrong by adding layers like 365 day returns or professional authentication. They didn't just capture a niche, they actually expanded it. They proved that a marketplace doesn't have to be everything to everyone. It's just has to be the absolute best at one thing.

[B]

C.J. i don't want to call you out too much right now, but do you want to show us your purchase history with StockX? I mean, how much. How much monthly revenue are contributing to StockX?

[A]

Yes, and that's why you should click all of our ad sponsors. Anywho, what does best at one thing actually look like? When it scales? StockX took the sneaker resale market and made it feel like placing a limit order, blind bidding, ask prices, a verification center that authenticated every pair before it got shipped. They turned a gray market into a transparent one, and the transparency itself actually grew the market. People who never would have bought a $400 shoe from a stranger on ebay were suddenly comfortable because the process felt professional. It wasn't like rolling the dice. Hipcamp did the same thing to camping. The problem wasn't that people didn't want to go camping, it's that getting a national park reservation had become the Hunger games. You're refreshing recreation.gov at 7am Six months in advance, hoping to snag a site at Yosemite like it's Ticketmaster. Not really a great experience.

[B]

And let the record say that the only thing stopping me from scaling Half Dome is that website. It's definitely not my favorite heights.

[A]

Hip Camp's insight was that there's a staggering amount of unused private land. You've got ranches, farms, vineyards, and they're owned by people who would happily let you pitch a tent for 40 bucks a night. That supply literally didn't exist as a category before Nobody was cold calling a rancher in Montana Ben asking to sleep in their meadows. Hipcamp created the supply class, and the demand followed. Reverb did it for musical instruments. If you've ever tried buying a vintage guitar on ebay, you know the problem. A $3,000 Les Paul described as good condition with two blurry photos and a no returns policy. It's hard to send someone money confidently. Reverb added real condition grading, historical price guides, and a community of people who actually understood the difference between a 59 reissue and a 72 original. They treated instruments the way StockX treated sneakers, like assets that deserved a proper transaction layer. Etsy actually acquired them, which tells you a lot about the niche they successfully professionalized. Okay, well, I got you, Ben. One more Fair is a sneaky one. The old model for independent retail buying was brutal. You have to fly to this trade show. You walk a convention floor for three days, you place orders based on gut feel, and you pray the inventory sells. Fair digitized the entire thing, which alone would have been useful but not transformative. The real wedge was the financial layer. Net 60 terms and free returns meant a boutique owner could take a chance on a new candle brand with basically zero downside. They de risk discovery for the buyer and unlock distribution for the seller. Hey, thanks for listening. We'll be right back after a word from our sponsors. Being a cfo, you know how much I love tools that actually make the lives of accounting and finance folks easier. One of my favorite tools right now is Rylit, the AI native ERP going head to head with netsuite. Yes, someone is finally doing it. I met Rylit two years ago when they were still in stealth. Since then they've absolutely taken the finance world by storm. Their mission is to make the zero day close a reality. And they're actually doing it. Customers are literally closing their books at 1:35pm on the first day of the month. They've got everything you need to scale your business. Complex revenue recognition, native integrations, custom reporting, multi entity close management, and much more. They're only a few years in and are already supporting NASDAQ publicly listed companies. Yes, seriously, if you want to scale your business on an ERP that wasn't built in the 90s, you need to check out Rillet. Book a [e-mail] CJ oh cool, cool. That's me. That's r I l-l e t.com CJ r I l l e t.com CJ tell him I sent you there. Ali and the team at Tabs are building something that directly addresses one of the biggest headaches I see finance teams deal with today. Pulling data from your erp, your CRM, your FPA tools and your usage systems and actually making it all work together. If you're running usage based pricing or complex contracts, you already know how this usually plays out. Contracts live in one system, usage data lives somewhere else. Billing happens in another tool. Revreck gets layered on top at month end and finance ends up just reconciling everything by hand to close the books. This is exactly the problem Tabs was built to solve. TABS is an AI native revenue platform built for controllers and CFOs. It brings together data from your ERP, CRM and real product usage into a single system of record. From there it automates billing, collections and revenue recognition without finance having to duct tape workflows together every month. If usage based revenue is core to your business, this is the future of how billing and Revrec get done. Go check them [e-mail] run that's tabs.com run. We've all been burned. We've all bought that enterprise planning tool that promised the world, only to realize six months later that we've basically taken on a second full time job just to keep the software running. I want to talk about a company that was built specifically to kill that cycle, Abacum. I actually remember my very first conversation with their founder and CEO Julio Martinez over three years ago. Back then they were just starting out in Spain. Fast forward to today. Julio moved the home base to NYC and they're the engine behind finance teams at Strava, Replit and JG Wentworth. Abacom doesn't turn you into a software admin. The integrations are actually self service. You don't need a $300 an hour consultant to plug in your ERP or HRIs. And they're doing AI in a way that actually impacts the things FP and A teams are doing every day. Things like creating variance summaries, building formulas and modeling scenarios. If you're scaling fast and you're trying to avoid that legacy platform trap, Abacom is the move. They're building the future of our tech stack and they're doing it with a CFO's perspective. Go to Abacum AI to see it for yourself. That is Abacum AI.

[B]

It seems that the pattern's always the same. I mean, you just pick one category that a generalist is serving and maybe serving poorly. You add that layer of trust, maybe a curation layer, a financial layer that the generalists can't justify building. It's just rinse and repeat. Do you think we're just in this perpetual cycle of bundling and bundling and bundling and unbundling?

[A]

I do think that like time is a flat circle and we just go from bundling and unbundling. But here with marketplaces, what they're doing is adding more value in the vertical playbook. It works. And what really made it work was when everyone got a supercomputer in their pocket. We've designed something wonderful for your hand. Many of the more recent marketplaces I mentioned wouldn't be possible or as attractive if they were only on a desktop. The iPhone launched in 2007. The App Store opened in 2008, and within a few years, three things that used to be impossible became trivial. You knew exactly where someone was, they could send you a photo instantly, and they were reachable at all times. Gps camera and push notifications. Boom goes the dynamite.

[B]

And yet no one has topped the app.

[A]

Ibear Mobile collapsed the time between intent and Transaction from hours to seconds. Uber is probably the purest example. The entire product is, you need a car, there's one four minutes away. Here's a map showing it moving towards you in real time. And then my wife talks to the driver. None of that works on a desktop. Same for Doordash. Same for Instacart. An entire generation of marketplaces didn't just benefit from mobile. They couldn't have existed without it. And mobile did something subtler to it. It actually lowered the barrier to becoming a seller. Listing a product on eBay in 2004 meant uploading photos from a digital camera, like with the stupid little USB thing in the chip, writing a description, setting auction parameters. It was a part time job. Then you have Poshmark and Mercari. They turn that into take a photo, set a price, post, boom, done. You could list a jacket on your couch during a commercial break. When you make selling that frictionless, you unlock a supply class of people who never would have bothered before. And suddenly you have a much bigger marketplace. But here's the thing about having a supercomputer in your pocket. It only has one home screen and most people only use four or five apps to do most of their buying.

[B]

And yet there's still only one app where I can virtually drink beer.

[A]

I spoke with VC Boris Wirtz of Version One Ventures and a former marketplace founder. He actually sold his business to Amazon. He told me what it takes to actually win in marketplaces over the long run. His answer surprised me. In the beginning, any marketplace that is starting to evolve, right, it's built around supply. And if you don't have to supply, you don't even have a marketplace, right? As the category grows, sellers on that marketplace will look for other opportunities to sell their inventory, sell their products. Supply, while necessary to differentiate at the start, is actually in eroding moat. What wins is mindshare, the buyer's habit, the muscle memory of which app you open first. You can find 99.9999% of the things on Amazon somewhere else. Their long tail of books got them off the ground. But they didn't win because of supply. They win because you're already on Amazon for 12 other things. So like, why would you go anywhere else? Uber and Lyft are the clearest examples. They started in the same category. The supply is literally identical. Most drivers, Ben, you probably noticed this, they're on both platforms. But Uber expanded into UberX Comfort Premium and then added Uber Eats on top of it. Lyft stayed focused and now When I need a ride, I open Uber without thinking. Not because the cars are any better, but because the app earned a slot in my mental rotation. And here's the mind blowing part. Over time, you aren't competing against other companies in your category. You're competing against all apps for Mindshare. Someone once said to me, you aren't competing against other CFO podcasts, you're competing against Taylor Swift. It was partly in Jess, but it's kind of true. All consumption competes for limited hours in the day. The average person has four or five apps that handle most of their commerce. If you're not one of them, you only get the leftovers.

[B]

All right, cj, I know you've been working hard on your album of covers, but I think it's. I think it's best to stick with just interviews and newsletters.

[A]

Well, in Asian markets, this has already played out to its logical conclusion. Apps like Grab, the Everyday Everything app handle deliveries, rides, and financial services all in one place. The end game for any marketplace is to become so embedded in your routine that switching feels like effort. Mindshare isn't just a marketing concept, it's the ultimate high ground.

[B]

Folks, you better break out your abacuses, your slide rulers, your TI89s, because I think CJ is about to drop some math on you.

[A]

Haha. And yes, it leads me to take rates. Ben, the main character in this story.

[B]

The vig, the rake, the juice, the slippery Steve.

[A]

I think you made up that last one, but yeah, yeah. So people talk about take rates as if they're just pulled out of a hat. But they're not. There are three characteristics I've identified that impact what you can, or rather should, charge. I make the distinction between can and should because they're absolutely examples of marketplaces who got greedy and just ran themselves under the ground. See Groupon. Bill Gurley wrote the canonical piece on getting too aggressive called a rake too far. So follow along with me. Take rate is a function of purchase frequency times ticket size times platform labor intensity.

[B]

I will hunting break it down for me.

[A]

Purchase frequency is the inverse relationship most people miss. The more often someone transacts on your platform, the less you can charge per transaction in absolute terms.

[B]

I feel like there's a nuance, though, between big rate on small transaction size and small rate on a massive purchase.

[A]

What are you, a quant or something?

[B]

Your what?

[A]

My quantitative ticket size works the same way in reverse. The larger the transaction, the lower the take rate needs to be. Nobody is paying a 20% fee on a $500,000 home sale. That's why real estate brokerages hover around 5 to 6%. But a $25 DoorDash order? You'll barely notice the 30% markup buried in the delivery fee and service charge.

[B]

I feel like we are extremely dangerously close to uncovering my doordash addiction. I definitely do not notice that.

[A]

Take rate well, payroll hits this Friday, then platform labor intensity is the big one. And it's the one that actually justifies higher rates. The more work you do, the more juice you get. Dan Hockenmeyer, who runs strategy and analytics at fair, framed this nicely in his newsletter. As a spectrum in the 10% range, you have the legacy. Amazon and ebay are essentially aggregating demand and generating leads. Thumbtack is even more honest about it. They charges a plumber a flat fee for the lead, somewhere between $10 and $100, and then they bounce. They know full well that the plumber is handing out a business card on the way out the door, and they built the business model to account for that disintermediation. In the 10% to 20% range, you've added trust on top of discovery. Etsy gives the basket weaver an identity, and the buyer of said basket a safe checkout. StockX physically verifies your sneakers before shipping them. You're not just introducing two strangers anymore. You're now vouching for the transaction. Okay, in the 20 to 30% range, you've built the entire logistics layer. That's hard doordash Uber Lyft. These companies employ or contract a fleet of humans who physically move products or people from point A to point B. That's not really a software margin business. That's true operational intensity, and their take rates have to reflect it. Vacasa sits here too. They're not just listing your beach house, they're unclogging your toilet at 2am When a guest calls.

[B]

That looks to me like an opportunity for someone to unbundle the plumbing side of the vacation rental business.

[A]

That's how we make our billions. Well, one thing that surprised me in the research is the supply side always pays. Every single marketplace I looked at charges the seller, which makes intuitive sense. The platform is bringing them business. The spicier debate is whether that business is net new or just offline demand that got rerouted or hijacked through a middleman who now takes a cut. More than half. 14 of the 25 marketplaces I studied charge both sides, and for some of them, the demand side fee more than doubles the effective take rate, only 3 charge the demand side more than the supply side. They are Airbnb, StubHub, and SeatGeek. And only one charge demand at least 3x more than supply Airbnb. Which brings us to a question worth asking.

[B]

When does a marketplace cross the line?

[A]

Well, to go back to our friend Bill Gurley, the core argument is counterintuitive. A higher rake is not always better. In fact, pricing too high might be the most dangerous strategic mistake a marketplace can make. The logic is simple. When you see it, your take rate becomes part of the landed price for the consumer. If you're charging 30% and your competitor is charging 10%, the exact same product costs more on your platform. You've essentially turned your own pricing into a reason for buyers to leave. And on the supply side, sellers, they're not stupid. They'll migrate to wherever the economics are better. High rakes don't just create friction. They create an incentive for someone to come and undercut you entirely. Gurley's favorite example was booking.com now, in the late 90s, Expedia and Travelocity were running what's called a merchant model. They were bundling vacation packages at take rates north of 30%. Booking.com came in with 10% and an agency model that gave hotels better economics and better cash flow terms.

[B]

All right, so what happened? What was the result?

[A]

Well, they signed up nearly every small hotel in Europe. More supply meant more selection for travelers, which meant more demand, which meant more hotels wanted it. And the flywheel spun because the rake was now low enough to let it. Now, here is the sexy part. Once Booking.com had dominant supply, they actually let hotels voluntarily bid up their rake for better placement. So the average rake climbed over time. It was market driven. It wasn't imposed. So hotels that wanted more visibility paid more. Hotels that didn't want that were punished by market forces. And when prices went up, the suppliers blamed their competition, not the platform.

[B]

Sounds like the same geniuses behind Google AdWords.

[A]

Wow, that's a good one. Exactly.

[B]

Yeah. Are there any other cautionary tales we got?

[A]

Of course. So Groupon was charging merchants roughly 38%, but that was after the merchant had already underwritten a 50% discount to the consumer. I used to work at a gym that did this. And if you do the math, the merchant was recovering about 30 cents on every dollar of value they delivered. So the effective rake was actually closer to 70%. That is not a partnership. That's more like a hostage situation. And merchants figured out pretty quickly that this was not going to be sustainable. Gurley also pointed at Apple and Facebook. They were both charging a flat 30% on their platforms. For Apple, that rake didn't just extract revenue, it repelled partners from using the App Store altogether. Amazon looked at the 30% cut on digital content and threw up on it. They decided they would just build a competing hardware ecosystem. The Kindle Fire exists, at least in part because Apple's rake made partnership math impossible. And Facebook's 30% cut on games eventually pushed Zynga and others to actively reduce their platform dependency. They started tracking percentage of revenue tied to Facebook as a metric to drive down, not up. Famous investor Peter Drucker called it one of the five deadly business sins, worship of high profit margins and premium pricing. As Gurley put it, there's a big difference between what you can extract and what you should extract. Water runs downhill. Hey, thanks for listening. We'll be right back after a word from our sponsors. Well, well, well. Here's what nobody tells you about being a CFO. You'll spend 50% of your time on stuff that is killing your momentum. The best CFOs I know are business leaders who know how to drive growth in heroic fashion. But most of us end up spending our days buried in manual work. I'm talking about collecting receipts, reviewing expenses and manually reconciling spend. It's painful. That's why CFOs need Brex. Brex built an intelligent finance platform that pairs corporate cards with built in expense management plus a team of AI agents to handle the manual finance tasks for you. That way CFOs have more time for the high impact projects that drive growth. You know the shit actually worthy of your CFO time. Bottom line, Brex is automating hundreds of thousands of hours of manual finance work every month across 35,000 companies like Anthropic, Coinbase and Doordash, ready to spend less time buried in expenses and more time driving results. Check out Brex app Brex.com metrics it is Brex.com metrics please guys. How the hell do I have three kids in daycare? Brex.com metrics you just launched a new AI product. The new pricing page looks great. I'm talking crisp but behind it. Last minute glued code, messy spreadsheets and running ad hoc queries to figure out what to bill. Customers get invoices they can't understand. Engineers are chasing billing bugs. Finance can't close the book. Looks well with Metronome. You hand it all off to the real time billing infrastructure that just works. Reliable, flexible and built to grow with you. They turn raw usage events into accurate invoices, give customers bills they actually understand and keep every team in sync in real time. Whether you're launching usage based pricing, managing enterprise contracts, or rolling out new AI services, Metronome does the heavy lifting so you can focus on your product, not your billing. That's why some of the fastest growing companies in the world, like OpenAI and Anthropic, run their billing on metronome. Visit metronome.com to learn more. That's metronome.com Here's a growth tax nobody talks about Every new monetization model you ship creates a nightmare for your finance team. Ad usage based pricing. Now you're tracking consumption against commitments. Launch product bundles. That's multiple performance obligations per contract. Offer mid cycle upgrades. Good luck reallocating revenue manually. But that's exactly where RightRev shines, right? Rev is the revenue recognition engine built for companies that can't afford to let accounting slow down growth. When your REV rack is automated, your product team can ship new pricing without asking finance for permission. And your sales team can close creative deals without worrying about downstream chaos. To get up on my CFO soapbox for a sec, I love talking about creative pricing models, hybrid pricing credits, tiered usage. But I've seen too many companies where the sales team is celebrating a huge quarter while finance is still trying to figure out how to recognize half of it. In a world where your pricing model might change three times next year, that flexibility is everything. If you want to scale your monetization without breaking your books, visit rytrev.com that's right, rev.com where modern monetization meets bulletproof accounting.

[B]

I feel like Airbnb kind of made this mistake too. I'm definitely trying to hoard my hotel points a little bit more now.

[A]

Well, let's talk about Airbnb, because for a while there, they were speedrunning every one of these mistakes. So this happened to me in my bachelor party. You'd find a listing for $150 a night. Great. You click through, you pick your dates, you hit reserve, and suddenly, boom. You stared at 287 bucks. Where did the other 137 come from? A $75 cleaning fee, a $42 service fee, an $18 Airbnb processing fee, which, I'm sorry, like what? What is that?

[B]

It feels. It feels so shady. I mean, like, I feel like I just want to pay for the thing that I'm paying for. Just show me. Show me the real price. Like why show me these other fees and just assume that. That they wouldn't be included. Or if they would be included. Like what is the processing fee? Processing. And the service fee is servicing. I don't know. Make it make sense.

[A]

And that's why it became a meme. People were posting side by side screenshots of the nightly rate versus the checkout total like it was a total gotcha. And honestly it was. The gap between the advertised price and the real price had gotten so absurd that people started doing the math and realizing they could just book a Marriott for less, which is the most damning sentence you can write about a disruptor that the thing you disrupted is now actually the better deal. The cleaning fees were the absolute worst offender. Hosts were charging $200 cleaning fees on a two night stay and then leaving a laminated note on the counter. Ben, I don't know if you've ever seen this, but make sure you strip the bed, start the laundry, take out the trash, and load the dishwasher before you check out.

[B]

Officer, I'd like to report a crime.

[A]

And remember Airbnb was only one of three marketplaces in my research that charge the demand side of more than the supply side. They were one of the only platforms where the buyer was paying a bigger cut than the seller, which is a bold choice when the buyer is also the one writing the review. To their credit, they've course corrected. In late 2022, Airbnb started pushing hosts to bake cleaning fees into the nightly rate and rolled out total price display so you could see the real number before you clicked the checkout chore list got publicly shamed enough that the culture shifted. The first stretch there. Airbnb was the poster child for what happens when a marketplace lets its fee structure become the product's biggest friction point. The platform that was supposed to make travel cheaper and more personal had somehow become more expensive and more annoying than a Holiday inn. You're not Dr. Stewart.

[B]

Well, I'm not a doctor, but I did stay at a Holiday Inn Express last night.

[A]

I set myself up for that one. So marketplaces charge a take rate, but the best ones figured out they don't have to stop there. I recently wrote a piece on what I call the Marketplace plus model. It's from speaking with Colin Gardner, who invests in marketplaces and was the chief product officer at Outdoorsy, an RV rental marketplace. The tldr is this. The smartest marketplaces provide something useful, usually software, to one side of the transaction as a hook to enter the network and Then they layer on additional revenue streams before, during, or after the core transaction. OpenTable is the cleanest example. Most people don't know this, but OpenTable started as a single player tool. Restaurants used it as a digital seating chart, a visual way to manage where parties would sit across time blocks. It'd replace pen and paper. That's it. There was no consumer facing product. There was no network. Just useful software for the supply side. And that wedge gave them an unfair advantage where they eventually opened the platform up and let consumers book online. By the time they turned on the demand side, they were already the system of record. The same playbook shows up everywhere. Outdoorsy built a SaaS product that turned a clipboard into a booking system for people renting out their RVs. They targeted the professionalized supply people who are already renting their RVs as a form of employment, but running the operation on paper. That cohort cared about utilization rates and maximizing revenue. If you give them real tools, they'll bring their best inventory to your platform, which means instant bookability for the consumer and better experience, which actually means more demand.

[B]

Ah, the unbundling of Craigslist has returned. We're back to the Gong Show.

[A]

True. And the plus doesn't have to come before the transaction either. It can come after. At PartsTech, where I was CFO, we served as a three way marketplace between auto parts manufacturers, suppliers and garages. Once a garage was already ordering parts on our platform, we offered them a SaaS tool to estimate labor hours for completing a job. That one addition tripled ARPU average revenue per user. We were making a percentage on each parts transaction plus a monthly SaaS fee. And it made us dramatically harder to rip out because we were embedded in their daily workflow. On the demand side, the most obvious example is Amazon. Pay an annual fee, get your stuff faster. And oh, by the way, there's a streaming library. The plus in prime turned a transactional relationship into a subscription relationship. Other examples, StockX charges for authentication. Airlines upsell trip insurance at checkout with the most apocalyptic language. My goodness. Click here if you don't want to protect your $782 flight to California during raging wildfire season. Okay, the pattern is always the same. Solve adjacent problems for your customers before, during, or after they use your core product. And every layer you add increases revenue per user and makes it harder for them to lead. The marketplace becomes less of a transaction platform and more of an operating system. Now I want to take you on a quick detour. Ben, are you game?

[B]

Let's do It.

[A]

Let's talk about every mother in law's favorite thing, Facebook Marketplace. Because I think it might be one of the most underappreciated businesses on the Internet and also somehow the most primitive. I'm going to throw some numbers at you. Facebook Marketplace has over 1.1 billion monthly active users. EBay has 134 million active buyers. And that's not a typo. Facebook Marketplace has roughly 8x the active user base of ebay. The total value of goods sold on Marketplace has been estimated at 98 billion. Ebay did 75 billion in GMV in 2024. So Facebook Marketplace is almost, I think it's certainly the largest peer to peer marketplace on the planet by volume and most people treat it like a feature, not a business.

[B]

So what's the deal? Why isn't the flywheel spinning here?

[A]

I don't know. Because the estimated revenue is somewhere around 30 billion. If they wanted it to be that eBay did 10.3 billion in 2024 and trades at a market cap of roughly 40 billion. If you were to slap a similar multiple on Marketplaces revenue, call it, I don't know, 2x4x, you're looking at a standalone entity worth between 60 and 120 billion. That would make it incredibly large. But it's a tab inside an app that most people open just to argue with their aunt about politics.

[B]

Okay, so maybe that's just a bit of an aggressive valuation.

[A]

Okay, well, hold my beer. It's worth at least 50 billion.

[B]

Okay. And yet the experience is essentially just. It's Craigslist but with your real name attached.

[A]

Exactly. Because there's no payment processing in most transactions, there's no shipping infrastructure for local deals, there's no authentication, no buyer protection worth mentioning, no condition grading. The transaction flow for the majority of Marketplace sales is just like, hey, meet me in this Wendy's parking lot. But I'm going to tell somebody before I go and hope I can come back. All right, so we spent the last, I don't know, minutes talking about how Marketplace is evolved by adding trust layers. They add verification, logistics and financial products. Facebook Marketplaces skipped all of that. Like I said, 100 billion in GMV running on profile pictures. In just good faith, the scam rate does reflect this. Over 62% of users report encountering fraudulent activity at some point. So maybe we're going backwards. Maybe Facebook Marketplace is just Craigslist 2.0. Kind of. But that's also what makes it fascinating from a business perspective. Craigslist proved that raw demand for local commerce is huge. Facebook proved that if you attach real identities to that demand, you get enough trust to transact at massive scale, even without bundling any of the infrastructure we've been describing. The question isn't whether Facebook Marketplace is sophisticated. It's whether meta ever decides to actually build on it. Because if they added some of the other bells and whistles, even just a fraction of what eBay or StockX has built, the 120 billion number that I said, it starts to look pretty conservative.

[B]

Okay, so other than getting kidnapped during a transaction or charging too high of a take rate, what are some other reasons that some of these marketplaces might fail?

[A]

Yeah, so not every marketplace works. I built one that didn't work. And the failures are often more instructive than the successes because they tend to share common DNA. The marketplace tried to become something it was never supposed to be. So the single most dangerous thing a marketplace can do is take on inventory. I cannot stress this enough. The moment you buy product and put it on your balance sheet, you're no longer a marketplace, you're a retailer. And retailers have a completely different risk profile, cost structure and margin, expectation and platforms. The whole beauty of a marketplace is that you facilitate the transaction without owning the thing being sold. You take a cut of someone else's risk. The second you start buying inventory, you've swapped a variable cost model for a fixed cost model. And you better pray that the market cooperates, because if it doesn't, you're holding a warehouse full of depreciating assets with nobody to sell them to.

[B]

But if you need to sell them to someone, I feel like there's. There's a good opportunity here to plug our history of EBITDA video that we did. So go check that out after this one.

[A]

Hodinkee is the case study everyone in media and commerce should memorize. They started as a watch content site, arguably the best watch content site ever built. Gorgeous photography, deep editorial, a passionate community of collectors who trusted Ben Kleiman's taste. The content was the product, and for a while, they monetized it brilliantly. They got affiliate commissions, limited edition collaborations with brands like Vacheron that sold out instantly, and eventually a watch insurance product that underwrote nearly a billion dollars in policies. That's a beautiful capital light business. Then they went bonkers and they decided to become a retailer. In 2017, Hodinkee launched the Hodinkee Shop, stocking new watches at MSRP from over 30 brands. In 2021, they acquired Crown and Caliber, a pre owned watch Dealer reportedly for about 40 million. This meant Hodinkee was now buying and holding inventory of pre owned luxury watches. Rolex's Patax aps. Right as the secondary market was peaking. The watch chart's overall market index hit 47,604 in March 2022. By January 2023 it was 34,569. That's a 27% drop. And it kept falling and falling and falling over 40% from peak to trough. Hodinkee was suddenly stuck with the double whammy. Every inventory holding business dreads a freeze in sales and plummeting asset values. The watchers were just sitting there in the crown and caliber vault and their worth less every month. The company went from 150 employees in September 2022 to around 30 after multiple rounds of layoffs across 2023 and 2024. The founder later admitted they hadn't been profitable for three years. The remaining inventory was eventually sold off to a New York based wholesaler. Could have sold it to me for a low seven figures on an acquisition that reportedly cost 40 million. Watches of Switzerland eventually acquired the media brand Sands inventory Sands retail operation. So the lesson isn't that watches are a bad business. They're probably not. StockX built a thriving watch and shoe marketplace without ever buying a single shoe. They authenticate, they facilitate, they take a cut, they never own the inventory. Hodinkee could have done the same thing. Run a consignment marketplace powered by the most trusted editorial brand in the category. Instead, they went from taking a percentage of someone else's risk to owning all the risk themselves. That's the difference between a platform and a store. And it costs them everything. Now Hodinkee isn't even the most expensive version of this mistake. Fab.com raised 336 million in venture capital, hit a $1 billion unicorn valuation, burned through 14 billion in cash a month and eventually sold for somewhere between 15 and 30 million. It was a total fire sale bet.

[B]

But CJ there has to be another way.

[A]

Of course there. There are a lot of ways to fail. The second way marketplaces fail is more subtle. And it's the disintermediation problem. That's just a fancy word for the buyer and seller cut you out. This is the fundamental risk of any service marketplace. Think about what happens when you hire a house cleaner through a platform. The first booking goes through the app, the cleaner shows up, they do great work, you're happy, and then the cleaner hands you a business card while they're walking out the door. Or you just exchange numbers. Why would either of you go back through the platform for the second booking? The cleaner gets to keep the 25% the platform is taking, and you get a lower price. The platform gets, well, nothing. Homejoy is the textbook example. Founded in 2012, they raised over 40 million, expanded to over 30 cities, and the growth looked incredible on paper. But only 15 to 20% of customers booked a second cleaning within a month. They'd use Groupon style promotions to acquire customers at a loss, and then those customers either churned or went directly to the cleaner. The unit economics were upside down. Negative lifetime value on acquisition spend that was already unsustainable. And you combine that with contractor classification lawsuits and inconsistent service quality. Homejoy actually shut down in 2015.

[B]

And I think Handy face a similar problem. Right, Right.

[A]

But Thumbtack figured this out early and designed around it. Instead of taking a percentage of ongoing transactions, which would create an enormous incentive for both sides to cut them out, they charged the service provider a flat lead gen fee upfront, Somewhere between ten and a hundred bucks per lead, depending on the job. That's it. They knew the plumber was going to hand out a business card. They knew the customer was going to call the plumber directly next time. So they stopped pretending they could prevent it and instead built a business model that captured value on the first transaction and let the rest go. It's a more honest model, and it's why Thumbtack is still live, while homejoy is now a footnote.

[B]

Okay, I'm starting to see a pattern here.

[A]

The pattern across all these Ben failure boils down to a simple diagnostic question. Does the platform add value to every transaction or only the first one? If you're Uber, you add value every single time. Real time, matching, routing, payment, insurance. You don't ever call the guy who picked you up three years ago to come and pick you up again. You just hit the Uber act. The supply is interchangeable, the demand is spontaneous, and neither side has an incentive to go around you. If you're a cleaning marketplace, you add value exactly once when you introduce the customer to the cleaner. And after that, you're just a toll booth.

[B]

All right, so what does the future hold here? What's the. What's the Cinnabon of the future that I get to sample now?

[A]

It all comes back to Cinnabon. Well, let's look forward. If someone told you 10 years ago that people would routinely spend $25,000 or $40,000 on a used car they had never sat in, never test driven, never Seen in person, that it would just show up on a flatbed truck in their driveway. You would have thought that was insane. Cars are one of the most personal, tactile purchases most people make. You're supposed to, no pun intended, kick the tires. You're supposed to take it around the block. You're supposed to smell the interior, make sure the previous owner wasn't smoking something in it. And yet Carvana sold 416,348 cars in 2024. Revenue hit 13.7 billion, up almost 30% year on year, they became, by their own accounting, the most profitable public automotive retailer in American history, as measured by EBITDA margin. They're targeting 3 million cars sold over the next five to 10 years. And that whole vending machine gimmick, it was cute. But what actually won was convenience. In the elimination of the dealership experience. It turns out people hated the negotiation, the pressure. The four hour Saturday at the fluorescent lit showroom. More than they needed to sit in the driver's seat.

[B]

So are we just going to buy

[A]

houses the same way it's already starting. Before the pandemic, roughly 3% of homebuyers purchased sight unseen. By 2025, the number is approaching 20%. The tools are getting better. You get the 3D walkthroughs, drone footage, AI powered valuations, remote notarization, digital escrow. A military family relocating from San Diego to Virginia doesn't have time to fly out for 12 showings. A remote worker moving from Brooklyn to Boise. I hope you're not doing that. Is scrolling Zillow the same way they scroll Amazon? But I think housing will be the last domino, not the next one. Cars are still standardized in a way that houses aren't. A 2022 Honda Civic is a 2022 Honda Civic, right? Regardless of which dealership it comes from. A three bedroom house in Denver could be wildly different from one two blocks away. The inspection matters, the neighborhood matters. And unlike a car, you can't return a house in seven days if the foundation cracks. The more likely near term future is that marketplaces chip away at the process without replacing it entirely. We'll search online, narrow down virtually, maybe make offers site unseen in competitive markets. But we'll still want to physically stick stand in the kitchen of the house we're about to spend half a million on. The marketplace will handle discovery, financing, title and closing. Everything except the gut check. And honestly, that's still a lot of value to capture, even without owning the final decision.

[B]

Now for the question that everyone's probably Wondering, how is AI changing marketplaces? What's going to change?

[A]

Well, I think the honest answer is that AI won't reinvent the marketplace model. It'll compress everything that's already happening and make it faster, cheaper and more personalized. On the discovery side, AI will replace browsing with matching. Right now you go to Amazon and search for running shoes and get 50,000 results sorted by some opaque combination of ad spend, reviews and relevance. I think in the near future you'll tell an AI agent I need trail running shoes for wide feet under 150 bucks and I hate the ones with thick soles. And it'll come back with three options. The browsing catalog becomes a two way combo. That changes the power dynamics, because if the AI is doing the filtering, the marketplace's ability to sell premium placement to sellers gets disrupted. Those ads make a lot of money. The billboard on the digital highway matters less when everyone has a personal chauffeur taking them directly to the store they need. On the supply side, AI collapses the cost of being a seller. Listing products, writing descriptions, pricing competitively, managing inventory, handling, customer service. All of that gets dramatically cheaper and easier. With AI tools, that means more supply enters the market, which is generally good for consumers, but creates more competition for existing sellers. The barriers to starting a small business on Etsy or Amazon go from low to nearly zero. On the matching side, especially for service marketplaces, AI could actually solve the disintermediation problem that killed homejoy. If an AI agent is continuously matching you with the best available cleaner based on your preferences, the cleaner schedule, and real time availability, the value of the platform now extends beyond the first introduction. The AI becomes the reason you stay. It's the same reason you keep using Spotify instead of just downloading the 20 albums you already like. The recommendation engine is the product.

[B]

All right, so what do I do about the fake CJ that's hitting me up for all these gift cards that he needs urgently? I mean, like the bad guys are using AI too. What do we do about those guys?

[A]

First, I did not ask you to get the gift card, so please don't run down in the store and do that. But it's a good question. On the trust side, AI powered fraud detection, automated verification, and reputation scoring will make marketplaces safer, which expands the addressable market. Every person who doesn't use Facebook Marketplace because they're afraid of getting scammed is a potential customer for a platform that can guarantee authenticity at scale. But here's what AI won't change. The fundamental structure of Commerce. Buyers need to find sellers. Sellers need to find buyers. Someone in the middle needs to facilitate trust, discovery and transaction mechanics. That's been true since the Mesopotamia, and it'll be true when AI agents are negotiating on our behalf. Which brings us back to where we started, Ben. The grain traders in Ur, the Persian bazaar, the Agora, Trajan's market, The strip mall in Yarmouth. Ebay. Amazon, that parking lot behind Wendy's.

[B]

We're back at Cinnabon, baby.

[A]

There's a popular myth in business that technology will eventually, quote, cut out the middleman. It's one of those phrases that sounds cute and revolutionary until you think about it for like more than 10 seconds. Because every time someone cuts out a middleman, a new middleman shows up, usually with a better app and a much better pitch deck. Amazon cut out the bookstore. Then Amazon became the bookstore. Uber cut out the taxi dispatcher, Then Uber became the taxi dispatcher. Airbnb cut out the hotel booking agent. Then they charged me way too many cleaning fees. But still, the middleman doesn't disappear. The middleman evolves because the middleman was never the problem. The middleman was always the solution to a coordination problem that both sides of the transaction would rather not deal with themselves. Sellers don't want to find buyers. Buyers don't want to vet sellers. Nobody wants to handle payment disputes. Nobody wants to build trust from scratch with every new transaction. That's the job. That's always been the job. What changes is who gets to be the middleman and how much value they have to add to justify their cut. The merchants in Trajan's Market paid rent to the state in exchange for foot traffic and security. Amazon sellers pay a 50% referral fee in exchange for 310 million active customers and and two day shipping infrastructure. The application has changed, but the premise is still the same. So if you're building a marketplace or investing in one, or even just trying to understand how commerce works, the whole thing distills down to a few principles that haven't changed in 5,000 years. Add more value than you extract. Don't take on inventory unless you're prepared to be a retailer. Build something that both sides need for every transaction, not just the first one. And remember, your real competition isn't another marketplace. It's the status quo of two people figuring it out on their own. You can't cut out the middleman. You can only become a better one. Ear on the Numbers is a mostly media production. Yelling and intro by Fat Joe Artwork by Meg delesandro show is executive produced by Ben Hillman. Nothing said on this podcast is intended to be business or investment advice. It's the sole opinion of me, a guy who feeds his dog way too much ice cream and has a history of net operating losses. Lol. If you like this podcast, hit subscribe and give us five stars. It will take like two seconds and our algorithm overlords love it. Drink water, call your mom and have a great day.

[B]

Peace.