May 30, 2024 · 1h 13m · in-depth
How to build and scale winning marketplaces | Casey Winters (Eventbrite, Pinterest, Grubhub)
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this in-depth interview with First Round Capital's Brett Berson, growth strategist Casey Winters breaks down the tactical mechanics of building, defending, and scaling venture-scale marketplaces, sharing frameworks on acquisition loops, supply quality, category expansion, and user activation.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Brett holds 16.9% of the talking time here. How this is scored →
Brett as informed peer 3.6 Guest teaching 6.1 Guest disagreement 1.2 Brett pushing back 0.8
speaking balance: gold is Brett, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 34:14 Copy high-frequency competitors immediately or die
Casey forcefully warns that marketplace founders face existential risk and will be completely wiped out if they do not immediately copy competitors who offer higher frequency or lower costs.
Hardest push from Brett ▶ 18:29 Brett challenges supply-side customer development assumptionsBrett directly pushes back against conventional wisdom, questioning whether supply-side discovery is essentially trivial since almost all merchants will take incremental demand.
Biggest teaching moment ▶ 21:08 The myth of the SaaS-to-marketplace transitionCasey rejects the common venture trope that SaaS businesses can smoothly transition into marketplaces, challenging anyone to name a single successful example from the last decade.
Brett holds their own ▶ 53:34 Brett deconstructs car-sharing unit economicsBrett demonstrates deep market insight by explaining how low absolute dollar returns fail to overcome the psychological friction of asset sharing.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Brett as informed peer | Guest teaching | Guest disagreement | Brett pushing back | Why |
|---|---|---|---|---|---|---|
| Core Requirements and Sustainable Acquisition Loops | 2 | 6 | 1 | 0 | Brett asks a broad opening question about the requirements for marketplace success. Casey delivers a comprehensive breakdown of supply fragmentation, demand promiscuity/frequency, and scalable acquisition loops. | |
| Evaluating Acquisition Loops and Growth Vectors | 3 | 6 | 1 | 1 | Brett asks whether founders discover acquisition loops early or mid-flight. Casey systematically explains why paid marketing degrades while scalable loops like SEO, content, and sales must be baked into product-market fit from the beginning. | |
| Modern Case Studies: Power and Fermat Commerce | 3 | 5 | 0 | 0 | Brett asks for fresh real-world examples of emerging marketplaces. Casey details how Power leverages search demand for clinical trials and Fermat Commerce uses DTC influencer networks for distribution. | |
| ZipTailor Concept: Marketplace vs. SaaS Network | 3 | 6 | 2 | 1 | Brett poses a hypothetical marketplace for tailors (ZipTailor). Casey reframes the prompt by distinguishing between true marketplaces (which own demand) and SaaS networks like Square or Shopify (which merely provide fulfillment). | |
| Non-Scalable Foundations and Liquidity Dynamics | 3 | 7 | 2 | 0 | Brett asks about early validation scaffolding for marketplaces versus software. Casey explains that software does not matter early on; only liquidity and manual execution count, citing Tony Xu personally delivering food in DoorDash's first year. | |
| Supply Side Dynamics and Workflow Bundling | 5 | 6 | 2 | 3 | Brett probes whether supply-side customer development is trivial since suppliers always want marginal customers. Casey pushes back with nuance, noting kitchens are underutilized fixed assets whereas front-of-house is constrained, and next-gen marketplaces must bundle workflow tools. | |
| The Friction of SaaS-to-Marketplace Transitions | 4 | 7 | 2 | 1 | Brett asks about the danger of getting trapped building a standalone supply tool without de-risking demand. Casey emphatically agrees, challenging the popular SaaS-to-marketplace thesis and recounting the immense cultural friction of transitioning Eventbrite. | |
| Eventbrite and the Challenge of Ephemeral Supply | 4 | 7 | 1 | 1 | Brett asks about cross-selling across disparate event types. Casey breaks down the core structural difficulty of ephemeral inventory at Eventbrite, and how low-capitalized public companies like Grubhub faced structural disruption from heavily funded models like DoorDash. | |
| Uber vs. Lyft: Execution and Price Dominance | 3 | 6 | 1 | 0 | Brett asks for Casey's perspective on Uber versus Lyft. Casey unpacks how Lyft's focus on friendly branding was superseded by Uber's price aggression and superior operational execution. | |
| Responding to High-Frequency Marketplace Threats | 4 | 6 | 2 | 1 | Brett notes how hard it is to update priors when facing new competitors. Casey cites Rover versus Wag, demonstrating that when a competitor aggregates the same supply with higher transaction frequency, a marketplace must copy the model immediately. | |
| Category Expansion Timing and Low-Frequency Models | 4 | 6 | 1 | 2 | Brett explores category expansion traps and whether strong fit in a niche translates into adjacent markets. Casey contrasts Whatnot and Goat, advising founders to focus entirely on winning the core category before attempting expansion. | |
| Low-Frequency Marketplaces: Zillow and Apartments.com | 4 | 6 | 0 | 0 | Brett asks about low-frequency marketplaces. Casey explains how Zillow uses non-transactional engagement (the Zestimate) while Apartments.com relies entirely on intense SEO optimization. | |
| Onboarding Supply: Active Training vs. Passive Assets | 4 | 6 | 1 | 1 | Brett highlights the challenge of maintaining quality when the real product is human labor. Casey explains how marketplaces must raise standards over time and contrasts active driver onboarding with passive asset monetization (Airbnb, Hipcamp). | |
| The Unit Economics of Peer-to-Peer Car Sharing | 6 | 5 | 1 | 2 | Brett articulates a detailed thesis on why peer-to-peer car sharing fails due to inadequate supplier compensation relative to friction. Casey agrees and expands with a financial bid-ask spread analogy, comparing Turo's daily model against Getaround's hourly model. | |
| Counteracting Cohort Degradation via Network Effects | 5 | 6 | 1 | 1 | Brett asks whether initial buyer/seller cohort behavior is fixed or malleable over time. Casey explains how cross-side network effects cause younger cohorts to perform better than older ones by outpacing the natural degradation of marginal users. | |
| The Activation Framework: Setup, Aha, and Habit | 3 | 6 | 1 | 0 | Brett asks if consumer activation frameworks apply directly to marketplaces. Casey reframes the concept into a three-part model: setup moment, aha moment, and habit moment, drawing clear contrasts between Grubhub and Pinterest. | |
| Custom Activation Metrics for Complex Marketplaces | 2 | 6 | 0 | 0 | Brett asks for additional examples illustrating the activation framework. Casey details the complex segmentation required at Eventbrite, where usage frequency varies drastically between weekly event hosts and annual festival organizers. | |
| Marketplace Failure Modes: Data, Promos, and Playbooks | 3 | 7 | 2 | 0 | Brett asks about common failure patterns. Casey outlines three major pitfalls: lack of data sophistication by geography/category, the trap of discounting to manufacture product-market fit (referencing Campusfood), and premature geographic expansion without a proven playbook. |
Statements from this episode (53)
Winters: Expedia and Booking make most revenue from hotels
“So you see with the Expedia's of the world, the bookings of the world, they actually make most of their money off hotels where there's a lot more variety and a lot more fragmentation.”
Winters: Low-frequency marketplaces require high AOV or high take rates
“If you're going to be low frequency, then you need to be high AOV. And if you can't be high OEOV, then you need to be high take rate on the low frequency. And if you can't get any of those to work, then chances are that a marketplace is not going to be an opti…”
Winters: Faire's zero-commission brand onboarding created proprietary acquisition loop
“Part of their unique advantage is they were able to build a product where brands can onboard their existing retailers for free and not pay any commission, but they can manage that workflow. And then that allowed FAIR to cross sell those boutiques to other bran…”
Winters: Edges in paid acquisition channels generally get competed away by rivals
“In today's environment where like paid is pretty hard to get an edge on any sort of edge you have will generally get competed away by competitors coming into the market. You're looking to see that someone's not going to rely on paid or sales for everything unl…”
Winters: Only four acquisition engines can scale startups to a billion dollars
“There aren't that many that helped you become a billion dollar company. It's basically sales, virality, what I call like content loops, which is we distribute content directly to, you know, Google or to social networks to bring more people in, and then those p…”
Winters: Paid acquisition inevitably degrades and becomes unprofitable over time
“And the challenge with paid is It always looks better yesterday than it does today, unless you have a network effect business where the product quality is getting better, faster than the quality of people you're bringing are getting worse, right? Cause you alw…”
Winters: Viral growth curves asymptote quickly as invited users exhaust interest
“Virality is, you know, mainly how consumer social companies have grown, but you get like a really fast curve, but then it asymptotes generally more quickly because all the people that reject the invites the first few times, they're not gonna necessarily accept…”
Winters: Fermat Commerce scaled B2B sales using central founder network nodes
“Part of how they got traction, mostly selling to direct consumer e-commerce companies. Is they found these nodes in the DTC e-com community that know all the other founders and they convinced those people that it was a great product. And so it's kind of like a…”
Winters: A true marketplace's core value to supply is bringing demand
“The difference between a marketplace and what I would call like a SaaS network is that a marketplace is primary value prop to the supply is that you bring the demand.”
Winters: SaaS networks yield fewer venture wins than demand-owning marketplaces
“It's generally lower take rate. If you have a take rate model, it's gonna be more like five percent versus 15%. And then of course you're paying the payment processing. It's harder to build a massive scale business. So you see fewer venture wins in that catego…”
Winters: Marketplaces win on liquidity and selection, not software features
“Marketplaces are not really usually winning on the software itself. Only if the software helps you find more selection on the demand side or brings you more money on the supply side.”
Winters: Tony Xu delivered DoorDash orders himself for the first year
“Tony at DoorDash was delivering the meals himself. For like the first year, right? Cause he didn't know how to acquire drivers yet. And he didn't want to work on that until he validated that there was a real product there.”
Winters: Restaurant kitchens are underutilized fixed assets constrained only by seating
“For Grubhub, that's basically never true because the kitchen is essentially always an underutilized fixed asset. They can always pump out more food. 99.9% of restaurants can pump out more food. It's the front of the house that gets constrained because they're …”
Winters: Next-gen B2B marketplaces must build workflow software to scale demand
“What you're seeing with a little bit more of the next generation of marketplaces is it requires more products to be in a position to drive demand scalably. Maybe you need to build out some workflow products. Or maybe you need to build out some sort of free man…”
Winters: SaaS-to-marketplace transitions almost never work as a second phase
“I generally advise founders against marketplaces phase two of the company. So there are a lot, a bunch of people were like, ah, SaaS to marketplace transitions. That's the model. And I'm like, name an example in the last 10 years where that's actually worked. …”
Winters: Marketplace network effects only kick in past 50% demand
“I don't think cross-site network effects really kick in until you're like past 50% of the demand, probably even more than that.”
Winters: Ephemeral inventory makes discovery matching very bad initially
“You have kind of ephemeral inventory, generally not a lot of data on the consumer side as to what consumer preferences are, because historically you haven't been collecting it. So then your ability to match And drive true discovery is initially very bad.”
Winters: Grubhub covered almost all US delivery restaurants under asset-light model
“Grubhub was an asset-like marketplace where the restaurants were in charge of doing their own delivery, and we basically had every restaurant across most of the U.S. That did their own delivery.”
Winters: Grubhub raised only $80M before its IPO
“So you're a company Grubhub that's raised eighty million dollars before you go public.”
Winters: The pandemic permanently fixed DoorDash's unit economics by eliminating latency
“Then the pandemic hits and all the unit economics of DoorDash flipped positive because everyone's all of a sudden ordering online and all the drivers are much more busy, which means all the latency in the model that was creating the losses went away.”
Winters: Grubhub's only winning move was acquiring DoorDash and appointing Tony Xu
“The only thing they could have done was buy DoorDash as early as possible, and then basically let Tony run the company and build a delivery network.”
Winters: A $100M public company cannot beat rivals with $4B in capital
“I just don't think there's a way that a public company that raised like a hundred million in IPO is going to outcompete. Companies that have raised four billion dollars each, I think, in Uber and DoorDash like to crush you.”
Casey Winters: Uber defeated Lyft by prioritizing low price over friendly branding
“Uber early on was all about legitimacy. These are licensed drivers. It's a black car, all this kind of stuff. And Lyft was like, oh yeah, it's like your neighbor and it's fun and it's friendly and it's safe. And then I think Uber was smart to say actually thei…”
Winters: Uber Eats improved driver acquisition efficiency by utilizing unqualified rideshare supply
“And it allowed them to use their supply that, that was signing up for Uber, but maybe didn't have a good enough car, or maybe the person was too young. So it allowed them to get some more efficiency on their supply acquisition as well.”
Winters: Uber Eats protected Uber from pandemic shocks that hit Lyft
“That made them a little bit less fragile to the pandemic because while, you know, rides shrank during the pandemic, food delivery demand went up and Lyft didn't really have that counterbalance only being focused on rides.”
Winters: Marketplaces must immediately copy rivals offering higher frequency or lower cost
“I think one thing marketplace founders need to be Careful about is if you are aggregating supply and demand in the market and someone else is aggregating the same supply and offering it at lower cost or with a higher frequency type of transaction, you need to …”
Winters: Rover would have failed if it had not copied Wag
“Rover copied the model and wag ran into some operational execution issues at scale and Rover eventually took the lead on dog walking as well. But if they hadn't, we'd forget they even existed.”
Winters: TAM is a misleading metric for early-stage marketplaces
“TAM is kind of a misleading metric for marketplaces, because in all the biggest marketplaces like Uber and Airbnb, it looked like there was no TAM.”
Winters: Whatnot scaled from Funko Pops into multi-billion-dollar livestreaming
“Whatnot is one of the companies I work with as an advisor, and their first category for their live streaming marketplace was Funko Pops, which is an incredibly tiny category, right? It showed that they could make obsessive collector communities hang out and tr…”
Winters: GOAT will probably stay sneakers-only, limiting its growth
“It probably always will be a sneaker marketplace, and that will limit ultimately how big that company could get.”
Winters: Most winning marketplaces remain single-category rather than multi-category
“As we look at most of the winners in marketplaces, you know, where I would say small winners from like a hundred million to a billion or bigger winners, which are a billion or more, most of them are single category, but then there's some really big outliers th…”
Winters: Zillow's Zestimate maintains homeowner engagement despite low real estate frequency
“I think Zillow has always been a really interesting one in that, you know, real estate's incredibly low frequency, but this estimate means everyone feels like they have a relationship with Zillow all the time if you are a homeowner. So it's a way for them to s…”
Winters: Low-frequency marketplaces typically remain fully reliant on Google SEO
“And I think that's normally what a low frequency marketplace looks is you're just grinding on SEO and conversion all day, every day. And there isn't really that opportunity to build that non-transactional product like Zillow has to create engagement. So essent…”
Winters: Grubhub data showed resolving bad orders boosted retention over flawless orders
“Early on, we were basically like, look, if there's an issue with the driver, if there's an issue with the order being late, we'll take it on as the marketplace, because we know if we make up for a bad experience by going above and beyond, we'll actually have b…”
Winters: Raising marketplace standards inevitably squeezes out individual sellers for power sellers
“The flip side of that is it creates a dynamic where only power sellers remain that can do all the standards. So the more, you know, individuals a lot of times can't keep up with the demands of the marketplace. And you certainly eBay is a prominent example of t…”
Winters: No single restaurant accounts for over 5% of Grubhub sales
“It's not like there's any restaurant at Grubhub that makes up more than five percent of sales. So it's still reasonably fragmented.”
Winters: Asset marketplaces win by requiring low initial supplier effort before raising standards
“So then there's the example of, okay, you actually have an asset that can be leveraged for the marketplace. And generally the way those models win, whether it's, you know, Airbnb or like Hipcamp is a company that I spent a lot of time working with. You're aski…”
Berson: Low rental payouts prevent peer-to-peer car sharing from scaling
“Even though someone has an asset that is not being utilized, it's hard to pay them enough to get them to actually move and care and give someone their keys and let someone take the car because they're getting nine dollars an hour or 15 dollars a day or 45 doll…”
Winters: Turo outperformed Getaround by prioritizing daily over hourly rentals
“I think Toro has done better than GetAround if I've paid attention correctly. And part of the reason why is GetAround built a bunch of infrastructure to make it easy for hourly rentals. And Toro is, wait a minute, there's no money in that. We should optimize e…”
Winters: Network effects are the only scalable defense against cohort degradation
“So what needs to happen is that the product needs to get better faster than the users you acquire get worse. And network effects are really the only scalable solution that can make that happen.”
Winters: Grubhub cohort retention improved year-over-year as restaurant supply scaled
“So what we found at Grubhub is that our retention got better every cohort because the product did get faster on the supply side, faster than the people we're targeting, you know, got worse.”
Winters: Pinterest doubled activation rates in one year by simplifying onboarding
“At Pinterest, we were able to double the activation rates after a year of work, but we had to dramatically simplify onboarding. We had to make sure it worked a lot better in all these international use cases that we weren't that good in. We had to delete a bun…”
Winters: User activation rate is the most important consumer metric
“And I believe in most consumer businesses, that is the most important number, which is how many you activate, meaning, and for those of you who are like, what does that mean? Of the people who sign up and try the product, how many have built a habit where they…”
Winters: Pinterest users pinning by week four reliably retained long-term
“For Pinterest, that was like, okay, we have to get them pinning at least once in the first four weeks. And after four weeks, if they're doing it by the fourth week, we know week five, week six, week seven, we got them. Like they're going to stick around basica…”
Winters: Grubhub users needed a second order in three days for retention
“For Grubhub, we had to get you a second order in the first three days to reliably predict you building that habit.”
Winters: Marketplace creator frequency depends on their business model, not software quality
“The frequency maps to their business model, not to how good our software is, right? The music festival person's not going to start putting on music festivals every week. Like, it's just not possible.”
Winters: Eventbrite measured only net dollar retention, not cohort retention, when he joined
“They actually, Eventbrite, when I joined, was not really in the habit of measuring cohort retention. It just measured net dollar retention.”
Winters: Marketplaces must build granular data capabilities faster than other models
“Marketplace businesses need to get sophisticated around data, usually a lot more quickly than other models.
And part of the reason for that is
Once you're doing more than one category or more than one city or neighborhood, the aggregate data doesn't really tel…”
Winters: Delivering two daily orders kept Grubhub restaurants from churning
“So at Grubhub, that was like, you needed to get them two orders a day to stick around as a restaurant, and then they would never, you know, churn.”
Winters: Discounting does not create product-market fit in marketplaces
“One is discounting to product market fit does not get you to product market fit.
So in a marketplace, you are selling people on the value of the selection, on the value of, you know, the service.
And if you then give a five dollar discount, then you start chan…”
Winters: Grubhub gave $10 mobile promos because app users doubled LTV
“So when our mobile app found product market fit at Grubhub, we decided to give everyone 10 dollars off their first mobile order because LTV doubled if you got the app.
And it was a little bit of a gambit, but a bunch of people used that promo, and then they st…”
Winters: Startups nearly went bankrupt blindly copying Uber's expensive launcher playbook
“A bunch of people tried to copy Uber's launcher strategy, and then a bunch of people like almost went bankrupt doing it because it's really expensive and they weren't raising billions.
You didn't really understand what your playbook required.
You were just kin…”
Winters: Owning demand matters most in marketplaces, contrary to Uber's philosophy
“And famously, Uber, like, does not believe this. They believe supply is all that matters, and I fundamentally disagree. Supply is It's a lot of times the tool to get demand, but owning demand is what keeps the supply happy.”