Frameworks for understanding marketplaces
|Erik Torenberg||Nov 23|| 4||1|
Note: I’m launching a private telegram channel to share links of things I read and listen to. Submit your email here and I’ll add you when I set it up.
Over the years, I've looked at many different marketplace businesses and have been interested in understanding how they work. Some of my favorite writing on the topic include Sarah Tavel, Kevin Kowk, and Bessemer's pieces, so I thought I'd take this week to unpack their ideas, others, and my own through this deep dive into marketplaces.
A marketplace, put simply, is a platform connecting two groups of people (or companies) that efficiently facilitates transactions. One side acts as the supply (Uber drivers, for example) and the other side, the demand (Uber riders). Generally speaking, the platform itself doesn’t own either side — a basic marketplace’s only goal is to facilitate transactions on the platform while removing the complexity from the customer’s view.
So why are marketplaces such good businesses? For starters, a good marketplace has network effects that bring high barriers to entry — once you solidify your place as a market leader, you're virtually untouchable. Also, marketplaces are inherently asset light, meaning you don't need to hold inventory, resulting in a cheaper business to operate, and easier pivots if need be. Airbnb is a great example of this — would you rather operate the Airbnb software platform, or a hotel management company? Think about the impact on each if a pipe bursts in a room — Airbnb removes one listing and loses the opportunity to make a service fee, but the hotel management company loses revenue and has to organize & pay for the repair. The other benefit of no inventory is the ease at which these companies can scale — just take Rover; you'd likely want to run this business over a national dog daycare.
When you compare marketplaces to pure-play e-commerce, the benefits to starting one become even more apparent. Overtime, marketplaces tend to have lower CAC, increased retention, and greater average order values. While it may be hard to get initial liquidity in the marketplace, it gets easier as network effects kick in on the supply and demand side. On the other hand, e-commerce is exactly the opposite — those businesses have to rely on economies of scale or adjacent markets to scale. In other words, marketplaces are harder to get off the ground, but easier to scale, while e-commerce businesses are easier to get off the ground, but harder to scale.
In the beginning, we had the listing era. Companies like Craigslist, Grubhub, and Yelp dominated the marketplace ecosystem essentially providing lead gen for other businesses and their end customers.
Then we saw the proliferation of transactional marketplaces like Thumbtack and Fiverr which facilitated transactions between individuals — it connected some vertically specialized individual with an end customer who could benefit from their service.
Then came managed marketplaces which integrated or owned supply, like Honor or Clutter — though capital intensive with lower margins and less scalable since you commoditize supply, they were more controllable and thus provided higher quality more predictable experiences than others.
The jury is still out on managed marketplaces. The bear case is that they’re capital intensive, have lower margins, are less scalable, and handicap network effects because you commoditize supply. The bull case is that it’s a more controlled & thus higher quality experience, and you can get away with lower margins if TAM is in the trillions (e.g. healthcare).
Horizontal vs Vertical
While horizontal marketplaces, like Amazon, target a broad range of customers (books, prescriptions, grocery and cars), vertical marketplaces pick a subset of the market (a specific vertical) to target while building an end-to-end experience that prior incumbents (often non-tech-based) could never replicate. For example, while you could buy event tickets on Craigslist or eBay, Stubhub is integrated with ticket providers so they could also verify your identity, show seating charts, and the like — the experience is clearly better and you’re not left empty-handed at the stadium gates.
As it relates to horizontal marketplaces, Sarah Tavel says the key is to, "Identify the verticals with low NPS & use product to leapfrog the liquidity of the incumbent’s supply base." Further, you want to "nail a small, overlooked and under-appreciated niche, then land & expand from there" (e.g Amazon). Horizontal marketplaces are harder to make work, but they can create much larger venture-scale opportunities than vertical marketplaces. 10 years from now, do you imagine people having 20 apps for different service categories on their phone? Probably not. We’ll probably have a couple super-apps that dominate the rest — the 1-2 apps we'll go to for food, transportation, delegating tasks, and more. So vertical marketplaces either likely have to expand and become horizontal, or get acquired. You’re seeing this cut-throat race today with UberEats, DoorDash, PostMates and GrubHub vying to become your super-app.
How to Build Marketplaces
Once you’ve picked a market and identified both your demand and supply sides, the first step in building a successful marketplace is to identify the "red hot center" — the group that feels the pain most acutely — because they'll jump through hoops to use the product initially. Once you’ve found that group, no matter how good your product is, you’ll have to find a solution to the chicken/egg problem; who’s going to use your product first when there’s no value without the other side of the equation?
Lenny Rachitsky’s research shows that the vast majority of unicorn marketplaces focused on tackling supply before demand. If you have an exclusive lock over supply, that can be used as a unique and compounding moat.
According to Kevin Kwok, one way to think about easing supplier acquisition is by leveraging an underutilized fixed asset. “The best way to think of underutilized fixed assets,” he says, “is as pure potential energy sitting in people’s homes, cars, and random tchotkes.” As we see today, Hipcamp leverages underutilized land, and Cloudkitchens leverages underutilized commercial kitchens; Airbnb uses homes, and Uber uses cars. With that potential energy, the asset can help early markets emerge for marketplaces.
The other great thing about underutilized fixed assets is that once you find them you’ve unlocked a latent supply that can act as a money printing machine. Since the asset is otherwise idle, if you can get the owner of that asset onto your platform (and repeat that process over and over again), you may be on your way to building a massive business.
So just like any other startup, locking in supply is about having a key value prop that the supply side cannot overlook.
To put it another way, paraphrasing Sarah Tavel, think about how you can make both sides of the marketplace happy — what does the supply side need that can be satiated by the demand side, and vice versa, and how does that make each customer happy as a result. “Building a marketplace,” Sarah says, “is a relentless pursuit of exceeding expectations to create happiness.”
In the beginning, especially to get your initial users, you’ll have to do a ton of things that don’t scale. Obviously you’ll have to be scrappy. Knock on doors, have 1:1 conversations, and onboard each individual user — do the things that will increase happiness, not just GMV.
Sarah says it best: “by pursuing happiness, you’ll achieve growth. But not vice-versa.”
“Much like MAUs is a vanity metric for consumer social companies, I believe GMV is a vanity metric for marketplaces … GMV does not get to the heart of whether you are creating enduring value or not. No matter how large an incumbent may be, they are always vulnerable to a new entrant that makes buyers and sellers happier. In other words, happiness — not scale — is your moat … If you think of growth as a means to increasing the average level of happiness per transaction (instead of the goal itself), it will focus you on quality growth over vanity growth.”
Happiness as your north star will help you create magic for your customers. It’s not the beautiful product UI that’s keeping people coming back, it’s the happiness (compared to that of your competitors) your well-crafted marketplace experience brought them.
Sander Daniels (founder of Thumbtack), advises to find the one thing that works (no matter what it is) on both the demand side and supply side, and then double down. For Thumbtack, email marketing to professionals and SEO for customers were the tactics that worked — he reported something like 98% of their early growth came from those single channels, so they went all in. He didn’t even consider competitors. Sander mentioned that if you’re going after a big market, it’s clear from the outset that there will be more 2-3+ winners in a space, depending on the market size. Also, if you’re devoted to building a marketplace, you’re signing up to play a really long game — it’s impossible to know who will be winning 5-10 yrs from now based on current metrics.
So to find a supply-side that fits the mold we’re looking for, seek out markets that have fragmented supply, but are also polygamous. Babysitting, for example, would not be a good sector because people in that market generally have monogamous relationships — families often have one babysitter, and their switching cost is incredibly high. (Not to mention they'd probably feel more comfortable with a peer's recommendation to who watches their children over an algorithmically generated one). On the other hand Uber taps into a supply of drivers, by leveraging their cars as the underutilized fixed asset, whose relationship to the rider is polygamous — as long as they’re background checked and their car is clean, it doesn’t matter who’s taking you from point A to B, which means users are less likely to go off platform with a specific driver.
When picking supply, you also want to ensure you're targeting a group with high frequency of use and high AOV (Average Order Value) — without frequent purchases and large enough order volumes, your GMV will suffer. Finally, as with any consumer facing product, you want a low friction sign-up flow, as well as a solid reason why users should stay on the platform. From there, you can then consider how to monetize.
Successful marketplaces all have monetized mainly through a take rate on GMV. As such, you want to target whichever side of your marketplace seems more inelastic, or less likely switch to a different platform. Why? Ideally, you want to recoup fully loaded CAC (on a net contribution margin basis) in the first 6 months of the business and grow your LTV to be at least 3x CAC in 18 months. Since measuring LTV in the early days is especially hard, honing in on the side that needs your marketplace most will be increasingly critical as you scale. Take Fresha as an example — they realized salons & spas were in need of an end-to-end SaaS product to run their operations, and as a result, they offered their B2B SaaS tool for free, thus attracting one end to the marketplace. Then all they had to do was wait for the suppliers' customer bases to convert to users of their platform — the suppliers already had the customer loyalty; with Fresha, the marketplace technology made the experience 10x better for all.
After you've figured out monetization, consider how you'll expand to different verticals. The main question to ask is: "Does this expansion decision lead to stronger product-market-fit, or am I expanding just because I'm supposed to?"
The issue is that most companies expand too fast — remember, nail product & retention before expanding — but you also don't want to wait too long. For example, Airbnb spent the perfect amount of time on their core business before expanding into experiences. While they could have launched "Airbnb for X" for dogs, boats, etc. they made the smarter decision to hone retention for their core business before branching out. On the other hand, Rover waited too long to expand, and are still in the same vertical in which they started.
With many companies taking the "[Unicorn Marketplace] for X" approach, it's important to ensure you don't make these common mistakes. The first mistake I see founders make is overvaluing their GMV growth while undervaluing net revenue & unit economics — again, retention (i.e. happiness) is what's going to keep your marketplace afloat, not just high GMV numbers. Another mistake I see is founders overvaluing growth at the expense of liquidity — instead of looking at just growth of users, consider how much value you're creating for each incremental user. How easy will it be for each user to be matched with another on the opposite side of the marketplace? How easily can supply fill the demand, or vice versa? Another mistake to avoid is relying on solely paid customer acquisition, and, as always, scaling too quickly.
The best way to understand common pitfalls is to look at case studies. Why do most "Uber for X" companies fail, for example? Well, for one, Uber has a 10x better experience, but it's also cheaper. Most Uber for X companies may be 10x better than their competition, but they were more expensive and also had low purchase frequency (ie. moving companies, dry cleaning, and the like). Or why didn't Homejoy work? Similarly, as a result of low purchase frequency, and low AOV per customer, users ended up going off-platform to develop monogamous relationships with cleaners. It was clear that once you found one you liked, you'd probably want to stick with them, resulting in Homejoy's downfall. This doesn’t happen with Uber because you won’t keep calling the same driver and blindly hope they’re only 2-3 minutes away. Connecting you reliably to a nearby driver was the value-prop, not just the driver themselves.
Evaluating Marketplaces As an Investor
“First, forget about the marketplace part. Start by looking at the business, like you evaluate any other business.
Most marketplaces fail not because of the marketplace, but because of more fundamental reasons — not building something people want, too small a market, inability to acquire users, etc. So start by looking at the bigger picture before focusing on whether this works as a marketplace.
Market, Team, Traction, Why now, do they have a unique insight, distribution strategy, moat, etc.”
Here are the questions to ask according to Lenny:
🤤 Demand Product-Market Fit: Do customers really want this?
Is this much cheaper or easier than the (non-marketplace) alternative? How badly do customers need this?
Can they maintain good unit economics while delivering a cheaper/better alternative?
Will they be able to aggregate all of the demand eventually?
🤑 Supply Product-Market Fit: Do suppliers really want this?
Does the supplier meaningfully benefit, e.g. significant income, high-quality demand? How badly do they need you?
How exclusive is the supply to this marketplace?
Are you leveraging an underutilized fixed asset?
✨ Scalability with quality: Can you grow while maintaining the magic?
Fill rate (e.g. % of times a customer finds what they want)
Positive experience rate (e.g. % of 5-star reviews)
How much “supply” even exists that can provide this service globally?
As it scales, will this supply be able to provide the same level of service?
How important is consistency and quality to the customer experience?
📈 High frequency and average order value: What’s the potential revenue?
The more frequent and the higher AOV, the better
If neither frequent nor high AOV, needs to be a huge market size, cheap user acquisition, and high retention
Is the business able to extract enough value (e.g. take rate) from each transaction?
🔒 Reasons to stay on-platform: Why won’t suppliers cheat?
What keeps supply or demand from taking this off-platform, and avoiding your fees. The answer is usually convenience, obfuscation, or protection.
How much more convenient is to stay on platform?
What protections or reputation do I build for staying on platform?
😏 Non-monogamous: Will customers need to come back?
Will customers need to come back for more?
Is one match/relationship enough? e.g. doctor vs. restaurant.
Does the platform get paid just once, or for the tenure of that relationship
🧩 Fragmentation: How hard is it for both sides to find each other today?
How fragmented is the supply and the demand currently? Could they easily find each other without your marketplace?
Are suppliers “leaning into” your platform? Are up and coming suppliers hungry to gain market share and compete?
What is the distribution of spend across suppliers? Do only a few matter?”
Seed (pre-launch): 1 on 5-6M post
Seed (150k/mo. GMV, 15% take rate) : 3 on 8M post
Series A (650k/mo. GMV): 7 on 25M post
Series B (2.5M/mo. GMV): 20 on 70M post
There’s a ton of whitespace for B2B marketplaces to spring up. Just look at these massive markets as examples:
Custom manufacturing ($1T)
Auto parts ($800B)
Home appliances ($500B)
Investment banking ($150B)
After 10+ years of consumer marketplaces, we're finally getting there with B2B marketplaces, partly because today's business owners are getting increasingly tech savvy. With a fragmented market of SMEs and domain expertise required to build a successful B2B marketplace, no wonder it took so long! Most B2B spend still remains offline due to the horizontal nature of the market (most don't address vertical specific workflows), as well as lack of integrated payments & lending within these companies. Also, it's much easier to digitize existing relationships rather than facilitate new ones.
However, with B2B marketplaces we can make it easier to create these new relationships among B2B companies and their customers. What changed? First off, there's the generational shift — millennials have higher expectations and want better products. Second, with the proliferation of APIs from companies like Stripe & Plaid, marketplaces can easily integrate lending & payments. Further, this API-driven architecture enables B2B marketplaces to build real-time multi-vendor product catalogs with accurate SKU and pricing information.
According to this great piece by Bessemer, there are three key opportunities in B2B purchasing.
The first is in wholesale marketplaces, which facilitate smaller, more frequent transactions for commodity products like apparel. A good wholesale marketplace enables price and vendor comparison for its users, which, in turn, streamlines offline ordering. Most of them take a freemium approach, offering a core workflow product for free and monetizing by selling ads or data, facilitating payments & lending, or by leveraging scale to negotiate volume discounts from suppliers.
The next opportunity is in high-friction marketplaces, which facilitate larger, less frequent transactions for things like used cars, freight, logistics, bulk commodities, and even investment banking advisory services. In a high-friction marketplace, SKUs are less standardized and comparable across vendors, which means a “managed marketplace” approach is often necessary to ensure quality customer experiences. In addition, successful high-friction marketplaces can charge higher direct transaction fees and see little to no churn, allowing GMV to scale quickly, making it ripe for venture investment.
In terms of b2b marketplaces, at Village Global we’ve already invested in the likes of Astorian, Shipwell, and Vori, and we’re excited to find even more budding B2B (and other) marketplaces to add to the portfolio.
Thanks to Vikram Rajagopalan and Zach Davidson for reviewing this piece.
Read of the week: The Rise and Triumph of The Modern Self
Watch of the week: Yuval Harari’s talk on new religions, old but good
Listen of the week: The Realignment with Anna Khachiyan
Until next week,