On Two-Sided Markets
Many technology companies can be viewed in the context of a two-sided marketplace (TSM), a special case of a multi-sided marketplace (MSM). This view has interesting implications for the dynamics of the business in addition to how product strategy and prioritization is decided. Having worked for over 15 years in TSM environments from Online Travel, to Affiliate Marketing at Zanox, mobile ad network madvertise, and many years in Financial Services, I thought it might be useful to provide a conceptual overview of the markets, dynamics, and implications. Since TSMs are so ubiquitous among internet economy success stories, we can also learn a lot from the likes of Google, eBay, Airbnb, Etsy, Kickstarter, oDesk, iTunes, Netflix, Spotify, Facebook, and others.
Any time you have two parties (often called buyer and seller) who are connected by some intermediary (often called a platform) you have a TSM. They are everywhere and due to their growth dynamics via network effects and the rise of a more connected and technologically-driven society, TSMs are especially powerful and high-potential businesses.
Some of these TSMs, since their content is organized and supplied partially or completely by hand are not exactly TSMs but rather curated marketplaces (CMs). Netflix is a good example of this because movie studios do not provide the supply themselves, but rather Netflix does it as a result of commercial license agreements. Often, the goal of a CM is to add tooling and maturity to the point where it can become a successful TSM. This allows for even more efficient growth of buyer and seller groups and lower operating costs for the platform.
When a TSM is first established, there is a classic chicken and egg problem. You need buyers to have a USP for sellers and vice versa, so which comes first? There are various ways to solve this problem. We will not address that here, since the growth dynamics after this initial stage are most important for us. When a TSM is first formed, it can rapidly rise to a dominant position, and the natural equilibrium state of the market is either a complete monopoly on the service provided or a shared platform arrangement with multiple options.
The dominant/monopoly approach requires a few things in order to be successful. First, it only works in markets where multi-homing costs, meaning the cost of participating in multiple platforms, is high for buyers and sellers. This makes sense, because if it were easy to use multiple platforms then buyers and sellers would constantly participate in multiple platforms.
In addition to high multi-homing costs, there must also be strong and positive effects for buyers and sellers for participating in the market. More buyers must attract more sellers and vice versa, forming a very strong positive feedback loop leading to the rapid growth of the market.
In relation to that, there must also be limited ability to differentiate. In other words, the market environment must be restrictive enough that it would not make sense to build additional platforms.
Clearly in the Online Travel space, the monopoly approach is not sufficient, primarily due to relatively low multi-homing costs, usually around billing information. As a buyer, using multiple platforms just isn’t very difficult.
In addition, a TSM which looks to be tending toward the monopoly situation is higher in risk due to the winner-take-all nature of that environment. The risk is high for many, and the reward is only high if you are the dominant player.
This option is different from the one above, largely due to underlying commoditization of many aspects of the platform. Credit cards are a good example here, with multiple large players (Visa, MasterCard, American Express, Discover, etc.) in the space. The reason is that multi-homing costs are extremely low and ability to differentiate is high (points, cash back, limits, and so on).
This is similar to the situation in Online Travel, where the underlying products are often interchangeable. Flights are a good example of this, as people are typically more sensitive to price than other things about a flight. The multi-homing costs of using multiple platforms are effectively zero, mostly just the time required to supply your billing and contact details to purchase tickets. Pricing is a critical component of TSMs as one side is often subsidized at the expense of the other, and we will now discuss the implications of having a subsidized side of the market.
The saying goes that you can’t serve two masters, but in a TSM that’s exactly the goal. Buyers and sellers both expect to benefit from participation in the market, and this often comes in the form of subsidized participation for one side. For example, when I was working with madvertise Mobile Advertising, a mobile ad network (which is a TSM between app developers and advertisers), the subsidized participant was the app developer. The level of subsidy was sometimes significant and for a while we even had 1 million Euros set aside which we spent on revenue sharing with app developers, solely for the purpose of attracting more supply into the TSM. In the online travel space, it’s the traveler who is the subsidized end of the TSM. Do you know of any online travel site which requires the user to pay in order to buy a ticket?
In addition to one side being a subsidized participant, it’s also normal that the same side is extremely price sensitive. If they do not like the cost, they will simply change platforms, which is easy because multi-homing cost is effectively nil. Since price sensitivity is such a large factor in the subsidized end of a TSM, this could cause one market to have lower prices which would lead to users changing markets, and a price war to erupt. This leads to the question of the difference between what buyers are willing to pay, sellers are willing to accept, and the fundamental topic of any TSM: liquidity.
Now that we’ve built up a bit of background, we can get the real meat of the dynamics of a TSM. In the context of a TSM, liquidity is the ability to efficiently match supply and demand with limited impact to the underlying price. In the online travel space, increased ability to buy a ticket for a given route at any time, without the transaction influencing the price of the ticket, implies higher liquidity in the market. Liquidity is the defining success metric of a TSM, and is the reason buyers and sellers participate in a TSM in the first place. If you couldn’t buy or sell what you wanted, or you could only do so with extreme swings in price, the market wouldn’t be very useful for you.
Keeping in mind that liquidity is the cornerstone of a TSM, and the subsidized end of the market is so extremely price sensitive, we can think about what more or less liquidity in a TSM would look like and how this liquidity changes the underlying price of the goods in the TSM.
Since sellers have a minimum price they are willing to accept for their good, and buyers want to pay as little as possible (recall they are the subsidized end of our TSM), the effect of increased liquidity is actually that the sale price becomes closer to the seller’s minimum acceptable price. In other words, margins for sellers go down to the point where it just barely makes financial sense for them to continue to operate. You could think of this as a kind of equilibrium price for the good in the TSM. There is really no way around this, other than for sellers to differentiate themselves, something hard to accomplish in a somewhat commoditized market, or for sellers to attempt to bypass the TSM entirely and form direct relationships with buyers. This latter situation, commonly referred to as disintermediation is to some extent a risk for all TSMs that do not continue to provide maximum possible liquidity.
In addition to the shrinking margin result that can arise in the absence of supplier differentiation, another effect of increased liquidity in TSMs is how prices react in changes to supply of goods and what the distribution of prices looks like from a long-term perspective. If you consider that sellers are competing with each other for buyers (which they always do because we never have perfect liquidity) then that means that sellers are forced to drop their prices in order to stay competitive (because the subsidized party in our TSM is extremely price sensitive). This is a kind of race to the bottom, with the result being that prices quickly reach their equilibrium state, which is the lowest price sellers are willing to accept. However, the process of reaching this baseline price is in direct proportion to the amount of liquidity in the market. In other words, the more liquidity the market has, the more uniform the prices become and the faster all prices adjust to a single sellers changing their price.
So if you want to have a highly-advanced TSM, that means you need as much liquidity as possible, which you can measure by how tightly prices for the same good are clustered in addition to how fast prices adjust. To use a concrete example in online travel, if every ticket for the same route costs more or less the same, and a change in price from one airline or OTA causes very fast price matching among others, the market is highly liquid.
So high liquidity means tightly-clustered prices and fast adjustments to price changes. Low liquidity means dispersed prices and slow adjustments to price changes.
In order to build a successful TSM, and increase liquidity as much as possible, the TSM must provide the tooling and environment in the marketplace to support buyers and sellers in finding the best matches. There are many examples of product improvements along these lines, and they generally involve providing increased information to both sides. This information efficiency, the cornerstone of the Efficient Market Hypothesis is closely related to liquidity, which we are trying to maximize.
Regardless of which approaches are taken in order to increase liquidity in the TSM, we must have some ways of measuring progress. This could be something as simple as looking at the price fluctuations for the same good in combination with how tightly clustered prices are for comparable items. This will reflect directly on the amount of liquidity in the market, and depending on what the distribution of prices looks like it could be a very revealing KPI for the business. Perhaps the only one which really matters in a TSM.