Marketplaces: The ‘Mongrel’ Needed to Start and Scale
--
If you haven’t yet read it, you can read my first article on the Marketplace Questions VC Will Ask here.
Marketplaces are a beautiful thing when they work. But make no mistake, they take ‘mongrel’ to start and scale.
You need to get to a point where buyers and sellers are contemporaneously pleased with each other and pleased with the marketplace experience you deliver.
You have multiple parties to please, parties that think you’re out to get them, and parties that will happily ditch you no matter how loyal you’ve been.
Ultimately, marketplace success is achieved in three overlapping phases:
1. attracting a critical mass of supply and demand;
2. making sure they keep coming back; and eventually,
3. cost-effectively trapping at least one side.
Most startups stumble at the first hurdle. Phase two hinges on whether you add incremental value each time parties come to transact, and phase three is about making it painful to transact anywhere else.
In this week’s article, I dive into the growth tactics and staying power of winning marketplaces. 👇
Phase One: Engaging Supply and Demand
To catalyse marketplace usage, you need to attract usage from supply or demand and keep them engaged until the other side arrives in critical mass. This is famously referred to as the ‘cold start’ or ‘chicken-and-egg’ problem, as it’s pretty hard to attract one side if the other isn’t there yet.
As mentioned in my previous article, most marketplaces target a particular geography or category in generating initial supply and demand. For each marketplace, there will be a notably harder side to attract, so it’s important to focus on a group of buyers and sellers that are highly likely to be interested in one another.
Geographical constraint makes sense where supply and demand need to physically meet. For example, when launching in initial markets, Uber had small-scale goals of getting a cluster of 30 drivers in a certain city to pick up riders within 15 minutes.
In most other circumstances, category constraint makes the most sense. Australian ticket resale marketplace Tixel focused on tickets for high volume, high stakes events (usually around shared holidays like New Year’s Eve) where people needed to make big decisions with high prices and high opportunity costs.
Once you’ve decided how you will constrain the type of supply and demand you will target, you need to nail onboarding the hard(er) side of the marketplace. ‘Hard’ is a euphemism for manual and laborious, and with finite resources, this process can only be so inefficient initially. Amongst top-performing marketplace operators interviewed by Lenny Rachitsky, 80% started with the supply side.
Attracting Supply
The difficulty of onboarding supply is a function of how much it costs a supplier to onboard relative to the reward that the supplier can foresee.
Here are some examples of how supply was onboarded in descending order of difficulty:
Company: Lyft (P2P) | Tactic: Piggy-backing
After Uber legitimised the market for P2P rideshare, Lyft was able to use venture capital to entice Uber drivers with targeted campaigns, referral incentives, and a delightful onboarding process. At the time, there was limited consequence for using multiple platforms as a driver, so getting usage was relatively easy.
Company: Faire (B2B) | Tactic: Building single-player mode
Building supply in industries that have existing business workflows can be more challenging. Faire, for example, struggled to build initial supply and had to offer retailer buyers attractive payment terms and free returns to reduce friction. Note that while this worked, it was exploited. It was only when it introduced credit limits on retailers, invoicing solutions, and ranking systems that supply was joining and having a good experience.
Company: Etsy (B2C) | Tactic: Pounding the pavement
Etsy, the online furniture marketplace, had to build its supply by physically approaching supplier shopfronts and stalls. It targeted key locations like Dumbo and Brooklyn in New York City, which had a burgeoning market for hipster and handmade furniture. It grew slowly through word-of-mouth for its first six years by steadily embedding itself in the community. It knew things were working when each buyer referred an average of four buyers onto the platform.
For more examples of crafty ways supply has been attracted, check out Lenny Rachitsky’s article.
Attracting Demand
In most cases, a focus on demand comes after engaging healthy supply.
Generating demand is easiest where the value of aggregated supply is making things obviously cheaper or more convenient.
In a conversation with Rampersand Venture Partner Hugh Williams who advised DoorDash for four years, Hugh shared the story of how DoorDash started its life as a few people by the phone at PaloAltoDelivery.com. In 2013, delivery services were not entirely novel, but almost no restaurants in Palo Alto did delivery. So, a group of students started fielding calls to deliver food around the Stanford campus. Very little software was used, but it was a 10x better consumer experience for busy students and teachers who would otherwise have to pick up their meals. It was going to be hard to scale, but it showed the service had a clear demand-market fit.
Demand generation can be more challenging when a market is nascent and customers are unfamiliar with a product or service. An example is carbon credit marketplace Pachama. To convince demand that credits listed for purchase were reliable, Pachama needed to partner with the top echelon of global brands like Amazon and Spotify to build its legitimacy and attract further demand. It is a clear example of a marketplace that needed to work closely with its initial supply and demand to educate and bring confidence before it could focus on scaling efficiently.
The most ideal and efficient way to attract demand is by word of mouth. If you want a live example of how this is being well catalysed, follow NZ-born sub-tenanting marketplace Kiki’s Instagram stories.
The below snapshot from Lenny’s article shows the ways well-known marketplaces initially attracted demand.
Without supply and demand both wanting to meet via the platform, you do not have a ‘marketplace’. Most startups actually cannot prove sufficient interest from both supply and demand — and fall over here.
But, if you do have supply and demand interested in using the platform to transact, your next challenge is proving your marketplace is on a path to positive unit economics and stickiness.
‘Positive unit economics’ means that you are making more money from customers in their lifetime compared to what it costs to get them using the marketplace. See Tim Fung’s article on this for more detail.
‘Stickiness’ refers to the degree to which a customer is motivated to return to the marketplace on their own volition.
Phase Two: Improving Experience Over Time
Achieving positive unit economics and stickiness comes from maintaining and ideally improving the experience marketplaces deliver as more transactions occur.
As more supply is satisfactorily compensated for something it offers, and demand can find what it’s looking for, the chances of them coming back or referring others increase. This is referred to as liquidity.
Liquidity
Liquidity is a close proxy for the extent to which you’ve made it as a marketplace. As marketplace liquidity improves, more supply and demand typically come to the marketplace organically. This signals future prospects of growth efficiency and the ability to stave off competition.
On the demand side, this is identified by an increasing proportion of searchers becoming purchasers, which is often labeled the ‘search-to-fill rate’. For a direct-to-consumer fashion marketplace like the Iconic, this is the percentage of sessions over a given period that result in a purchase.
On the supply side, this is identified by the extent to which supply is delivering at its capacity via the marketplace. For local holiday van marketplace Camplify, a decreasing vacancy rate of the assets listed on the platform signals a positive trend for supply, while the % of times people who visit the website make a booking is positive on the demand side.
The ease with which liquidity is achieved also depends on how easy it is for a transaction to be fulfilled. Julia Morrongiello neatly separates marketplaces into three categories: double opt-in, buyer picks, and marketplace picks.
Liquidity is hardest when both sides must agree (ie Airtasker), easier when only a buyer picks (ie the Iconic), and easiest when the marketplace chooses for you (ie Uber). Marketplaces often face a trade-off between giving participants decision-making power and automating the search/match process. Therefore, tracking participant satisfaction alongside liquidity is important.
Flywheel Effects
Liquidity shows signs of growth efficiency, but what’s also magical about marketplaces is how ensuing flywheel and network effects entrench marketplace dominance.
Positive flywheel effects occur when the marketplace experience improves through a series of reinforcing actions. A positive flywheel effect demonstrates a marketplace’s ability to fuel its growth, and in some cases sustain growth with limited need to focus on user acquisition.
Below is a simple flywheel effect that Expert360 benefits from.
Expert360 facilitates the delivery of consulting services to business customers. Through ratings, reviews, and company environment data, Expert360- uses AI to improves its matching capabilities. Business customers have a great experience, and recognise that independent consulting is a great gig. They then become independent consultants themselves, and Expert360 adds to its catalogue of quality supply.
However, not all flywheels have solely positive impacts, and marketplace startups need to test whether adding certain functionality compounds or reduces flywheel effects in aggregate. For instance, Etsy introduced a community forum feature that drove great engagement for its existing community, but this also saw some sellers in specific niche markets find their pool of regular buyers and take them off the platform.
Network Effects
Network effects are a subset of flywheel effects that specifically involve the improvement of a marketplace as more demand and supply join. An important thing to categorise is whether network effects are localised (more concentration in a certain area improves the experience) or global (it doesn’t matter where people come from, just that there are more of them).
In my first article, I noted that not all startups are born equal. Sameer Singh notes that ‘Tier-1’ marketplaces are surfacing differentiated supply and are serving cross-border markets. Generally speaking, marketplaces where the supply is undifferentiated and locationally constrained face limitations with regard to their potential for efficient growth and market capture. Sameer Singh classifies these as ‘Tier-3’ marketplaces.
Taken from Sameer Singh’s article on marketplace liquidity
Therefore, to be successful as a Tier-3 marketplace, you need people repeatedly coming back to transact on their own accord. The amount of times someone returns has a high impact on your lifetime value relative to customer acquisition cost and payback period (which determines the extent to which you can reinvest cash in growth). Grocery delivery marketplace Instacart has succeeded here because repeat usage is so high.
Another potential limitation for marketplaces with undifferentiated supply is whether the marketplace does indeed improve with more volume. A range of fifty Airbnbs of different sizes, aesthetics, locations, and prices to choose from is much better than one. But for Uber, as long as a car is available to pick you up in five minutes, you don’t care if there is one or fifty available. This can frustrate the supply side, which is why Uber uses Rampersand-backed, PredictHQ, a real-world event data platform that helps it incentivise drivers to go to areas where demand is predicted to outstrip supply.
Phase Three: Becoming Indispensable
Any good marketplace is definitely going to have some fast followers and copycats. So, marketplaces need to use customer usage over time as a means to stave off competition. One way is introducing systems of trust and another is building significant value for participants pre and/or post-transaction.
Trust
A constant challenge for a marketplace of any size is building and maintaining trust between buyers and sellers.
Craigslist is a classified ads website that famously became the place where you could search for supply of just about any product or service. While this was valuable for parties to discover each other, it did not provide the workflows suitable to build sufficient trust and repute between counterparties for a swathe of different transaction types and value chains.
What it did do, however, was pave the way for many vertical-specific (and some horizontal) marketplaces to emerge.
To bring comfort to transacting parties as they scale, marketplaces that emerged after Craigslist often include:
- A rating or review system — this is fairly intuitive, but a track record of good behaviour bodes well for marketplace usage. In the case of two-sided marketplaces like Airbnb and Uber, both parties are kept to account by a review system, which compounds its value.
- Verified supply — As mentioned in my first article, Expert360 only allows 1 in 10 consultants to sell their services on its marketplace. The process of onboarding consultants is necessarily manual to ensure that the talent on their platform meets a standard of excellence.
- A safety net — promising parties a certain level of goods or service delivery is an impactful but often expensive instrument for securing trust. Perhaps the most exploited supply guarantee of the last five years is UberEats’ customer satisfaction policies. People have been able to repeatedly complain and get a full refund for food that was adequately supplied. Fortunately for Uber, the distribution power it has allows it to compel restaurants to bear the costs of this.
- Social proof — as Anand Iyer points out in an article on liquidity, marketplaces are social businesses. They grow through entrenching dependence on the marketplace to build track record and social signal. Testimonials, badges of achievement, and endorsements from influencers can go a long way to building brand reputation and repeat usage.
Dependence
As mentioned in my previous article, take rates are a proxy for how much marketplaces reduce transaction costs.
Over time, take rates become increasingly a function of ‘switching costs’ relative to ‘transaction costs’ saved.
A switching cost comprises the time effort required to have the same level of transaction experience using a different marketplace.
Particularly for the supply side, if you use the marketplace to manage your inventory or demonstrate track record for example, and there is no fluid way to list/sell/share elsewhere with the same distribution capabilities, you might avoid list/sell/sharing elsewhere at all costs.
How you achieve liquidity, flywheel effects, and overall stickiness will determine the difference between creating a generational marketplace and joining the marketplace graveyard.
Measuring this over time and identifying the crucial metrics for your marketplace is key to decision-making and capital-raising success.
Stay tuned for my next article — a comprehensive guide to Marketplace Metrics.
If you’d like to chat further about this article or how Rampersand thinks about marketplaces, get in touch. My email is hunter@rampersand.com.
