What is Marketplace Liquidity?
Liquidity is one of the most important things, if not THE most important, to ensure the success of a marketplace. We usually deal with two-sided marketplaces, for which we have:
- Demand (side 1): those who are looking to buy a service or a product from other users or companies
- Supply (side 2): those who are looking to sell a service or a product to users
A nice definition of liquidity for a two-sided marketplace is the following; the probability of selling something you list, or of finding something you are looking for (James Currier, NFX).
As the definition shows, you need to take into account both sides of the marketplace: users are able to find what they are looking for in your platform and sellers are able to sell their products or services to someone. Balancing the two is hard but vital: too many users without enough products to find, and those users will churn (abandon your platform) and find what they are looking for elsewhere, too many products without enough users, and it will be the supply who will churn to find other channels to sell through. This is yet another example of the “chicken and egg” problem, where it is unclear what is more important to get first. Usually, though, it is the user side that is harder and more important to obtain, as having users who are active and with high intent to buy will easily attract sellers.
Before jumping into how to measure liquidity, let’s first take a look at important characteristics of the marketplace to consider.
Which characteristics matter when dealing with liquidity?
- Number of sides in the marketplace
I mentioned before how you’ll need to measure liquidity at each side of the marketplace, so complexity will grow if the number of sides in your marketplace increases.
Take for example a delivery app, which is usually considered a three-sided marketplace with users (demand), restaurants (supply) and drivers. In this case, an additional metric for drivers liquidity is needed, for example “% of active drivers with an assigned order at a given time”. Similarly to the other liquidity risks, if this percentage is too low, it will mean that drivers are not earning enough per hour through the platform, if this percentage is too high, it could mean the service gets saturated and deliveries need to be canceled or delayed.
- Location constraints
There are marketplaces that are constrained by locations and distances. Take for example a mobility app: users are only able to get to “close enough” destinations, drivers will be offered rides that are “close enough” from their position… As a result, liquidity will change depending on the geographical area considered.
For marketplaces that are not so much constrained by geographies, measuring liquidity at a global level might suffice. This geographic flexibility acts also in favor to facilitate a proper balance between supply and demand, as any buyer can access almost any seller in the platform and vice versa.
- Number of product categories
Product categories can also have an impact on liquidity. If users in the platform are only willing to convert to very specific categories, understanding liquidity for each category will also be a must. An example where this is important is in job boards: users will only convert to job positions similar to what they are already working on or that match their career goals and expected benefits. It is very unlikely that a “Product Manager” applies to a “Software Developer” position, and if he or she did, the company would probably not consider that application. Balance (enough supply and demand) needs to be achieved for each category.
Other marketplaces have better luck and are usually defined as having “interchangeable supply”. Drivers in a delivery app would meet that definition, as the user doesn’t choose or worry about which driver brings the food as long as the food arrives at their doorway.
- Type of commitment in the marketplace
On one side we have single-commit marketplaces, where one side of the marketplace suffices to close the transaction. In the Buyer-Picks model, it is the user who buys the product from the supplier, without the involvement of the other side (e.g. e-commerce). In the Supplier-Picks model, it is the supplier that chooses an available job to fulfill (e.g. mobility app drivers). From a liquidity perspective these models are great as transactions happen fast and almost frictionless.
On the other side we have double-commit marketplaces (e.g. rental portals), which require both sides of the marketplace to commit to the transaction. These marketplaces have a harder time optimizing and measuring liquidity, as the final transaction doesn’t necessarily happen in the platform. Weaker signals are usually used as proxies, for example, the rental portal might ask a renter why is he or she canceling an apartment listing (“were you able to find your new tenant in our platform?”).
So, how do we measure liquidity?
There are metrics that combined wisely and taking into account the characteristics of the marketplace listed above, can give a good enough sense of how your marketplace is doing in terms of liquidity. Let’s consider a typical two-sided marketplace, we will need metrics to measure the two sides.
Buyer liquidity: probability that a user finds a product to convert to.
- Vacation rental portals: percentage of sessions started by users that end up booking at least one accommodation.
- Delivery app: percentage of sessions started by users that end up ordering through the app.
Supplier liquidity: probability that a product listed is booked or bought by a user:
- Vacation rental portals: percentage of accommodations that are booked daily
- Delivery app: percentage of restaurants that get at least an order daily
Supplier liquidity is specially complicated as it has a dependency on the timeframe used to measure (while this doesn’t happen from the demand part as it can be measured at session level). Take the vacation rental example, “percentage of accommodations that are booked daily”, here the timespan is set to 1 day, but this could be too small:
- You could expand to a timespan of 1 week, and compute something like “% of nights with booking out of the total nights availability for a given week”. This could be better from a variability perspective, but homes with 0 nights booked would compensate with those with more than one night booked.
- An alternative could be “% of homes with at least 1 booking for a given week”, ensuring this way your demand is better distributed through your supply. But is 1 booking a week really enough to keep a home satisfied with the platform? Would that minimum number of nights required vary based on the home type?
- I worked on a last alternative to solve these pitfalls when measuring supply liquidity during my time at Glovo (a delivery app). The proposed “fairness metric” was based on a statistical measure of dispersion (Gini coefficient), to understand how well orders were distributed across the restaurants of the delivery app.
Together with this summary of marketplace liquidity, I’d like to share my most important learning: there is no perfect single metric.
The reason: it is not possible to summarize something as complex as liquidity through a single number. My advice is to track not only several liquidity metrics adapted to the marketplace characteristics, but also other general metrics such as user activity (MAU, DAU…), revenue (GMV, margin…), relationship between supply and demand (buyer seller ratio or overlap) or user satisfaction (NPS, churn, LTV…).
Resources:
