Marketplaces (WIP)
The key problem for a multi-sided marketplace is the chicken and egg problem. Should demand come first, or should supply come first?
The answer, with some nuance, is supply. A marketplace needs to have something to sell. That means you’re now in the business of manufacturing liquidity. Eli Chait has described some of the most successful strategies for doing this:
Create value for the supply-side before demand arrives: Chris Dixon calls this “come for the tools, stay for the network”. What software, data, services, etc. can you provide to engage suppliers before demand even exists?
There are some risks here. I imagine it’s particularly difficult to move from tool provider to marketplace if you’re selling your tool and generating revenue. Eli gives the example of OpenTable as a successful marketplace, but my sense of that business is that they struggled to shift from being a restaurant management and CRM platform to being a true marketplace and generating revenue from that model. The incentives are entirely different: restaurants want to own their their customers and their data; diners want a bunch of different options. I would guess this is why you see a lot of higher-end restaurants who can generate their own demand moving away from OpenTable.
Generate incremental servicable demand: Most businesses (particularly services businesses) have a problem with variability in demand. At certain times of the day/week/month/year they’ll be at peak capacity, while the rest of the time they’ll be below (often well below) capacity. Providing either off-peak or high-value demand to these businesses is very valuable. It is often also nearly pure profit, because there is a high proportion of fixed cost in these businesses that any incremental demand (revenue) helps to cover.
There are risks here as well. Incremental demand often has some combination of lower revenue/higher cost to acquire than existing demand. And unlike the value creation example above which has some lock-in, you’re likely to be in a very difficult battle to efficiently acquire supply against well funded competition.
Build a marketplace where supply and demand significantly overlap. If your demand is also a significant part of your supply, you can grow by targeting a single group. This seems to work well for niches and collectables, where both sides of any transaction could just as easily have swapped roles (an early hit for Ebay was beanie babies). A good argument for starting narrow and building from there.
The obvious risk here seems to be that if the overlap is less significant than you think it is, your spending is going to be very inefficient, and you’re dead. Also, you may be forever constrained to your niche, which may be quite small (Bezos chose to start with books because while it was niche, there were 3 million books in print at any given time).
Andrei Brasoveanu has put together a good list of marketplace metrics here. A few takeaways:
Marketplaces have a different CAC dynamic, where they are acquiring not just demand but also supply, and any other players in the marketplace (if it’s a multi-sided marketplace). Forgetting to account for CAC on all of these is ☠️
Average order value will go some way to determining what channel mix you use for distribution. But so will the aggregate order value of any individual buyer or seller
If there is a high average order value/relatively long cycle to a sale, what are the other measures of engagement you can track on your platform? Andrei suggests messages, but there are lots of other potential indicators
Pay attention to how concentrated your marketplace is (or isn’t) becoming. At some stages a concentrated marketplace is necessary. At some stages it significantly increases the risk of ☠️
You should know at all times what distribution channels are ready to be scaled, and which are inefficient for one reason or another. When you need to turn the crank, where should the money go?
There are also a lot of marketplace dynamics that need to be taken into account, which vary from marketplace to marketplace:
How differentiated is supply? Where does the marketplace sit on the spectrum from every item it supplies being exactly the same (Uber, roughly) to every item it supplies being entirely different (Airbnb)?
Does a purchase take supply off of the market (Airbnb) or is supply effectively infinite (Kindle books)? If it takes supply off the market, for how long? How frequent are purchases as a result? Is utilisation high for individual goods/services, but low in aggregate?
Does the cost/quality level being supplied by the marketplace match what is being demanded by users? Are you matched/mismatched on other characteristics of supply (category, for example)
How important is geography to your marketplace? Is it relatively agnostic (Upwork) or almost entirely driven by location (Tinder)? What is your marketplace density?
Resources
https://andrewchen.co/marketplace-startups-best-essays/
https://blog.elichait.com/2018/04/09/how-the-100-largest-marketplaces-solve-the-chicken-and-egg-problem/
https://cdixon.org/2015/01/31/come-for-the-tool-stay-for-the-network
https://twitter.com/JonErlichman/status/1189233985798463488?s=20
https://medium.com/@algovc/10-marketplace-kpis-that-matter-22e0fd2d2779
https://techcrunch.com/2017/07/11/marketplace-liquidity/