What makes a marketplace work?

What makes a marketplace work?

Marketplaces have been a keen area of interest at 1984 Ventures. Over the past 5 years, we’ve invested in several marketplaces — including B2C models like Kyte, Streetfair, Teachme.to and Fay Nutrition and B2B platforms like Trusted Health, Seso Labor and Exo Freight. Informed by these investments, we’d like to share our framework on what makes a marketplace work.

There are five criteria in our framework. While the first four may appear somewhat formulaic, the fifth criteria - execution and ingenuity - is equally important. Without out-of-the-box thinking, a new marketplace idea may not be novel enough to stand out from the crowd (and may never overcome the chicken-and-egg problem).

1. The marketplace’s solution must be significantly better, faster or cheaper than the status quo

A key learning has been that a marketplace needs to be significantly better than the status quo (”10x better”) in order to gain (and maintain) traction. This could be in the form of it being more convenient, substantially cheaper or altogether novel compared to what exists today. And technology can play a big role in delivering this improvement. Some examples of 10x better alternatives below:

  • Kyte (Uber for car rental) -- having a rental car delivered to your door is significantly more convenient than spending hours lining up at an Enterprise or Hertz store
  • Fay Nutrition (marketplace connecting consumers to nutritionists) -- by doing the hard work of partnering with insurance, Fay drastically reduces the cost of consulting a nutritionist (10x cheaper), which builds virality and gets the flywheel spinning
  • Swimply (AirBnB for pools) -- being able to access pools in your neighborhood is something you never previously never had access to, similar to how staying at someone’s home was inaccessible before AirBnB

Some Swimply pools in your neighborhood for a hot summer day

2. Consider the strength of the network effect

Investors like marketplaces because they can exhibit network effects at scale, making them difficult to displace (e.g. Whatsapp). But network effects are nuanced and their strength varies on a case-by-case basis. This is because at a certain scale, an incremental unit of supply doesn’t necessarily enhance the value of the platform.

Adding differentiated supply to your platform tends to correlate with the strongest network effects. It tends to be easier to add differentiated supply of goods (e.g. Airbnb, Faire) than labor (e.g Upwork), but needs to be assessed on a case-by-case basis.

Some examples below (in ascending order of the strength of network effects):

  • Upwork -- customers go to Upwork to hire knowledge workers (e.g. software engineers) for temporary gigs. But if there are 1000 vetted software engineers on Upwork versus 500 on another platform, is Upwork necessarily better? You can argue that beyond a certain threshold of supply (which can be a surprisingly low number), both offerings are comparable
  • Uber -- by definition, Uber's supply is undifferentiated - which means that it can be more easily copied by competitors. Once there are enough drivers on the platform in a particular city, the wait time for a car is low and adding more drivers in that geography doesn't meaningfully improve the customer experience.
  • Doordash -- keeping factors like price and reliability equal, customers in a particular city will choose the food delivery platform that has the restaurants they like (Marafuku Ramen!) as well as the breadth of cuisines (Burmese). Adding the 20th Greek restaurant in a city doesn’t improve the customer experience. For a rival platform to get to parity in a city would require having the best restaurants as well as the breadth of offering, which while doable, is not a particularly easy feat
  • Etsy -- goods marketplaces like Etsy theoretically benefit from strong network effects because every new unit of supply can make the platform more interesting to the customer. But if the incremental supply is undifferentiated (e.g. the 500th ceramic mug), then customers also feel burdened by infinite choice
  • Airbnb -- every new AirBnB added in a particular city enhances the value of the platform for customers because the supply is differentiated and gives the customer more choice and optionality. You can argue that after a point, too much choice hinders rather than helps the customer. But, this would likely only happen once you have substantial depth in a market - and can be solved through technology (e.g building thoughtful filtering tools that serve as a benefit to the customer)

It’s important to note that network effects typically only exist once the company gets to scale. At the earlier stages of company building, founders need to do what it takes to overcome the chicken-and-egg problem. This has been written about extensively (see here, here, here), but typically you need to think creatively to find a distribution advantage which helps crack one side of the marketplace (e.g. paying drivers to be on the Uber/Lyft networks in the early days, or creating restaurant management software to attract supply e.g. Opentable, or scraping/faking supply to give the perception of a thriving network e.g. Thumbtack).

3. Aim for high frequency, high fragmentation and avoid monogamy

Every marketplace founder should read Bill Gurley’s canonical post on marketplaces. Bill calls out a few elements that we also look for:

  • High frequencya low frequency marketplace means that you need to intercept the customer at exactly the right time in order for them to transact on your platform. In contrast, high frequency categories (food, rideshare) have a better shot at building a repeat customer relationship and increasing LTV
    • Low frequency marketplaces often lack a strong network effect, which makes customer acquisition and the associated unit economics extremely critical.
      • For example, car selling is a low frequency category and isn’t naturally prone towards a strong network effect (i.e. more sellers attract more buyers, but more buyers doesn’t necessarily attract more sellers). As such, one car selling platform may not be substantially better than another and both likely compete for the seller, driving up CACs. However, it’s still possible to have a healthy marketplace in this category (e.g. Motorway) since the high sale price of cars can more than cover CAC
    • Low frequency marketplaces can still succeed by being well positioned to capture demand when it arises (e.g. have a dominant SEO position when someone searches), or building additional products to increase touchpoints with the customer (e.g. Zillow with the Zestimate) - h/t Casey Winters
  • Highly fragmented supply it’s difficult to intermediate a market where the supply is concentrated. For example, OTAs like Priceline and Expedia have been unable to make money on air travel (given airline concentration), but have had success with hotel bookings (since hotels are extremely fragmented)
  • Avoid monogamy! — certain categories are more prone to disintermediation once a match is satisfied. For example, a marketplace connecting customers to hairdressers will trend towards monogamy: once you find a hairdresser you like, you won’t return to the platform.
    • However, various “monogamous” marketplaces have been able to continually intermediate transactions after a match is made by thoughtfully crafting their business model (e.g. Fay Nutrition’s nutritionist marketplace encourages repeat interactions with a nutritionist, naturally increasing LTV; similarly, Cambly uses a subscription model to monetize language tutoring once a pupil has selected a tutor (see grey box below)
    • Other marketplaces build software to empower the service provider going forward, which can also keep the transactions on platform (e.g. Future.fit provides its personal trainers with scheduling software to manage their burgeoning customer base, as well as building in post-session training videos for the customer to keep them on the app).

English language tutoring marketplace Cambly provides an interesting case study as it relates to monogamy. Language lessons is a category that would normally trend towards monogamy, as once you find a good tutor you don’t have a reason to return to the marketplace. In response, Cambly built tools that try and keep you on the platform (e.g. a virtual classroom + scheduling software), and have also opted for a subscription model where you sign up for a series of classes (versus just paying for individual lessons). This requires customers to prepay for 10 lessons, which offsets disintermediation risk and increases LTV

4. Ability to extract strong unit economics

So you’ve come up with a great marketplace concept that’s better than the status quo, trends towards a network effect and doesn’t suffer from low frequency, high concentration and monogamy. Now you need to demonstrate that the unit economics can be favorable, namely, that you:

  • Have a high LTV
    • Thing like frequency of use tend to increase LTV, whilst monogamic tendencies tend to limit it. However, as mentioned above, there are ways you can build product or thoughtfully construct a business model to increase LTV, even when your category is naturally limited by monogamy and frequency
    • Additionally, high AOVs that substantially cover CAC can also make the unit economics attractive for low frequency marketplaces (e.g. car sale marketplaces like Cardino.de)
    • Many marketplaces are able to use their wedge product as a wedge to also offer SaaS solutions, giving them opportunities to monetize further. An example from our portfolio is Seso Labor, an agricultural labor marketplace connecting farmworkers to farms. The company also sells workforce management software to farms and has more recently moved into managing the workforce’s payroll as well, which substantially increases the revenue per customer
  • Cover both demand CAC and supplier CAC (an additional cost that doesn’t exist in SaaS)
    • Without a clear network effect, CACs can increase as the marketplace grows, as incremental customers are more costly to acquire. Having strategies to minimize paid CAC and/or maintaining a healthy differential with LTV are critical
    • Note that supplier CAC is an additional cost that doesn’t exist in SaaS
  • Cover operating costs once at scale (such costs can become significant if you have a more managed marketplace approach)
  • Have an efficient CAC payback period (<6 months is great)

An interesting case study of failed unit economics is on-demand shipping startup Shyp, previously a Silicon Valley darling. Shyp’s model was for customers to take a photo of their item, leave it on their doorstep and have a courier pick it up. The courier would transport the item to the company’s HQ, at which point a delivery service like Fedex or UPS would get it to the end customer. However Shyp overestimated the frequency with which customers would ship items and also their willingness to pay (Shyp was charging $5 for courier pick up and $3 for packaging). Shyp eventually pivoted to focus on SMBs rather than consumers, but still couldn’t get the unit economics to work.

5. Execution and ingenuity

When envisioning their idea for a marketplace, we encourage founders to think creatively about how they can reimagine the unit of supply, any hacks for solving the chicken-and-the-egg problem, or creative ways to price the product. After all, the world is an efficient place and most obvious marketplace ideas are already taken - but a new idea that captures the imagination can completely change the perception of what was previously considered a solved (or unsolvable!) problem.

Some examples of out-of-the-box thinking below:

  • Hipcamp (Airbnb for camping) — by activating a previously inaccessible unit of supply (people’s backyards and properties), Hipcamp added completely new inventory to the mix, generating “found revenue” for suppliers
  • Teachme.To (marketplace for sports lessons) — instead of focusing on professional sports coaches, Teachme.To targeted talented former athletes who were easier to acquire and around whom visually appealing content could be built, which would rank higher in SEO. Additionally, instead of focusing on dominating one category/geography before moving onto the next, the founders discovered a way to launch a new city for a new sport each day, radically altering the speed at which the business could grow
  • Kindred (members-only house swap network) — coming out of the pandemic, many knowledge workers were working remotely from various Airbnbs across the country. Kindred observed this phenomenon to create their own house swapping network. For a subscription fee, people could contribute their house to the network (say, an apartment in San Francisco) and get access to other people’s homes for short stays (say, a cabin in Lake Tahoe). House swapping has informally been around for decades, but a productized solution with a clear tailwind propelled Kindred’s early success


The ability to spend a week working from a cool house in another part of the country at a fraction of the cost of an AirBnB is a compelling value prop.

The final (and more gloomy) point to note is that while this post contains a lot of examples of successful marketplaces, there’s a lot to learn from the failed attempts as well. An interesting repository of failed marketplaces and their post mortems can be found here.

On a more positive note, if you’re building a marketplace that you think satisfies many of the criteria above, I’d love to hear from you! Feel free to email me at samit@1984.vc.