Welcome! The Marketplace Theory series on MarketMaker deep-dives into the different characteristics of marketplaces, and how these differences lead to different marketplace dynamics, and require different product interventions.

1:1 versus Many:1 matches

Most classic marketplace products match parties one-to-one - one buyer is matched to one seller, one rider is matched to one driver, one homeowner is matched to one handyman, etc. Even a group booking an Airbnb is 1:1 - that group of guests is a single buying “party”, looking for a single place to stay. 

However, you might be running a marketplace that has many-to-one matches, which results in a more “chunky” rather than liquid marketplace:

Examples of chunky marketplaces include:

• OpenTable, where multiple dining parties may be matched to one restaurant

• Eventbrite, where multiple guest parties may be matched to one event

• Shared rides on Uber or Lyft, where two or more riders are matched to a driver

• Wonolo, where 50 or more workers may be matched to one shift at a warehouse

Chunky marketplaces can introduce a lot of new (and horribly complex) marketplace dynamics.

Complications

We’ll cover some of these complications today, using an imaginary startup as an example. Rentalot is an “Airbnb for parking spaces” startup (you rent… a… parking lot…).

(Like all good startups, Rentalot was born from its founder’s personal experience. A few years ago, Ruth had a coworker who owned a truly gorgeous sports car, which he was street parking in SF (#bayareaproblems). He was going to be out of town for a few weeks and didn’t want to leave his car alone. Ruth, however, was in possession of a parking spot in her building’s garage, but also possessed a deep and abiding hatred of car ownership. A deal was struck, and in exchange for a sushi lunch and getting to ride in said gorgeous sports car, the car got to chill in Ruth’s garage for a few weeks, and the car lived happily ever after.)

Now, like all good marketplace startups, Rentalot needed to get big fast on both sides now, and as such, struck up a deal with commercial parking lots in SF to quickly bring on a lot of supply.

Rentalot’s SF supply profile now looks like this:

Which is quite a classically chunky profile.

Complication 1: Distortion of inventory

The first complication that chunky marketplaces have is distortion of inventory. When a car owner comes to Rentalot and views our available spaces, they’ll see just 15 unique search results because we only show one result for each location, but we actually have 55 open spots:

This can accidentally make your marketplace look more empty or thin than it actually is, and may reduce trust in your platform.

For Rentalot, we can solve this quite easily by adding some kind of signal about how much inventory is actually available, for example, by showing the number of spaces at each location:

But it’s not always desirable to show true inventory for all marketplaces. Think carefully about your user’s needs and your business needs when deciding how much to disclose.

Complication 2: Denominator problems

For this second complication, I’m gonna get a bit mathy on you. Let’s say Rentalot is trying to sign a fourth commercial parking lot, and that operator, quite sensibly, asks us what the occupancy rate is.

A funny thing happens when inventory units (e.g. a parking garage) have different sizes - when you take rates or percentages, all of that gets condensed. If we take the average of all occupancy rates per property on Rentalot, we get 78.4% (wow!). But, if we took a simple sum of how many spots were booked versus how many spots we had available on the platform we get… 49.1% (oof). So which number are we going to use when we pitch to this commercial lot?

This can drastically complicate how you compute key business metrics and KPIs. 

As your marketplace gets more sophisticated and you start A/B testing, this will also have huge ramifications there as well, as chunky units can unbalance the randomization and lead to false negatives or false positives. Any A/B test that uses occupancy rate as the goal metric will overweight occupancy changes to single-lot locations, and underweight the change to high-lot-size locations.

I also… Don’t quite have a suggested solution to this. (So ideas and thoughts are very welcome!)

In non-perishable marketplaces like eBay and Amazon, this problem seems to be avoided by simply doing measurements on the non-chunky side. For example, eBay may run a buyer-randomized A/B test with conversion rate as the goal metric (buyers here being the less chunky side). We can be reasonably certain that that increases revenue, without having to look at some kind of “percent of inventory sold”. (It’s also unclear that you want to sell 100% of inventory for some marketplaces. Either inventory can be replenished, which makes it a weird moving target, or you actively want to avoid a stock-out.)

But, in perishable marketplaces like OpenTable and Wonolo, this kind of occupancy rate or fill rate can be a key metric. A restaurant will naturally want to know if you managed to get them 10 out of 10 tables booked per day, or a warehouse wants to know that if they asked you for 50 workers, are they getting 15 workers or 45. For the other marketplace folks out there with chunky marketplaces and denominator issues, I’d be super curious to know how you handled it.

Complication 3: Concentration risk

If one side of your marketplace is dominated by a few high-demand or high-supply parties, that generates concentration risk.

Let’s say that it’s currently the annual worst week in San Francisco - the Dreamforce conference, when all traffic patterns get royally borked. (Fine, fine, Bay to Breakers is worse, but it’s just one day, and realistically a parking lot startup should worry much more about Dreamforce than a road running race.)

The operator of Market Parking calls us up, and says “sorry, I need all my spaces for Dreamforce!”... and you’ve just lost 55% of your available supply, probably right when you needed it most. 

To be fair Dreamforce isn’t a great example here as it’s a macro-level effect, and we might expect even some individual lot owners to pull supply out during that event. But this dynamic still stands - your internal champion at the customer might quit, that client may shut down a product line, or a freak tornado levels the warehouse.

Diversifying and de-concentrating both sides of your marketplace will help counteract this risk, and also reduces your risk of being disintermediated (where the parties deal with each other directly and cut you out of the deal).

So there you go, 3 weird complications that a chunky marketplace may bring. Curious if folks have seen others!

Thanks for reading! Subscribe below to make sure you get each new article from MarketMaker. I also welcome comments, corrections, and feedback - connect directly with me on LinkedIn for that.