Defensibility in digital platforms
The main driver of long-term value creation
Samaipata is an early stage founders’ fund investing in digital platforms, displaying increasing returns at scale, across Europe.
“Would you invest in a marketplace in the absence of network effects?”
… A tricky question I was recently asked by David Pritchard, a marketplace founder (OpenTable Europe) at an event we, at Samaipata, organised in London for marketplace founders. It took me a while to figure out my mind, but you can find below an extended version of my answer alongside some of the comments made that day. Keep in mind that, as always, opinions are my own!
Defensibility in Digital Platforms
Network effects (NFX) are a crucial piece for the defensibility of digital marketplaces.
NFX act as an economic moat, the business’ ability to protect its long-term profits and market share from competing firms. The vast majority of the value will come from long-term defensibility.
NFX are definitely one of the most powerful moats but are not the only source, let’s analyze the main sources of moats and their importance.
There are 3 Moats that we usually see in marketplaces, and each of them can make a difference:
- Network effects (NFX)
- Brand equity
- Switching costs
Economic literature refers to a few others, such as economies of scale, patents, etc. while relevant, I don’t usually see them making a difference on their own when analysed against the competition.
1. Network effects (NFX)
Although it’s a concept that is widely used these days, few people understand how complex and different in nature NFX can be. There are a few classifications that help cast some light into them and understand how powerful they can be as a source of defensibility.
NFX occur when the value of a company’s service increases for both new and existing users as more people use the service. As a quick reminder, network effects and viral growth are very different things. Viral growth has to do with the ability to acquire users at almost zero cost. One can classify NFXs in different ways.
1.1 Different types of NFXs depending on the number of users involved
As per the graph above:
(1) Positive linear NFX: is the ideal situation for a marketplace, where NFX are most valuable. Regardless of the scale, the value of the platform keeps increasing with each new user and/or usage. e.g., used car marketplace. The higher the number of sellers and buyers the better. A variation of this type could be an exponential curve where each new user adds more value than the previous one. Airbnb is a very good example here. The more cities and accommodations per city, the better for a demand-side user (guest). The more users the better for supply side user (host).
(2) Positive- linear from a tipping point (minimum liquidity): Another variation of the previous one, a very widespread one, is the case when the value perceived by the user remains low and flat-ish until certain critical mass is reached (x), AKA minimum liquidity.
(3) Asymptotic NFX: Above x number of users the value perceived by users in the platform remains constant. The lower the “x” the less powerful the NFX as a way of defensibility. To calculate that tipping point, in La Nevera Roja, we plotted the number of restaurants serving in a given location and the conversion rate for thousands of searches in our web and we were able to draw the regression line that followed the exact asymptotic pattern. In our case from 0 to ~ 45 restaurants in a 7’ bike ride catchment area the conversion rate increased exponentially and then reached a plateau. There were other different NFX in place, but the impact of adding new restaurants tended to zero at that point, because users didn’t find any value in the marginal restaurant added, representing clearly asymptotic hyperlocal NFX.
(4) No NFX: represents a business without any NFX, the value of the platform remains constant regardless of the number of users. Example: a traditional one-sided e-commerce model.
(5) Negative NFX: where each new users detracts value from the marketplace. e.g. a very exclusive VIP group that loses its value as it becomes mainstream.
1.2 Direct and indirect NFXs
1.2.1 Direct: increase in use drives direct increase in value to all other users in the platform. It was formulated for the first time by Robert Metcalfe related to Ethernet and is known as Metcalfe’s law. The law states that the effect of a telecommunications network is proportional to the square of the number of connected users of the system. It’s the simplest one of the NFX. e.g.: the value of a telephone or a messaging service (Whatsapp) is proportional to the number of users I can communicate with.
1.2.2 Indirect: increase in usage of a service results in the production of new products or services that increase the value of the original product. e.g.: the more people have an iPhone the more developers will build apps for the App Store, therefore, the value of the iPhone will increase. Salesforce AppExchange is another good example that follows the same rational.
The main difference about Direct vs Indirect is time. While Direct NFX increase the value of the platform almost instantaneously, indirect NFX will require time to kick in.
1.3 One side vs. multi-side NFXs (who benefits from the new users joining the network)
1.3.1 One side NFX: most of the times one side of the marketplace leads to an increasing value for the opposite side of the marketplace and negative NFX for the same-side of it. e.g.: An employer in Cornerjob will benefit from the addition of a new job seeker to the platform but will suffer slightly negative NFX from another employer joining the platform.
1.3.2 Multi-side NFX: occurs when every user in the platform benefit directly of the fact of a new user joining the platform. Social networks, messaging apps, etc. show this kind of NFXs.
The same thing happens in marketplaces where supply-side users can operate as demand-side users and vice versa. Additionally, those marketplaces are usually cheaper to scale as you are getting two for the price of one when acquiring. e.g: at 21buttons an influencer (supply-side user #1) can be a buyer (demand-side user) and vice versa. So that, when acquiring a user, you could be acquiring a micro-influencer and vice versa. Airbnb is another nice example of this.
1.4 Local vs. global NFXs
1.4.1 Local and Hyperlocal NFX: occurs when only users in a given geographical area benefit from a user joining the platform. Note that in existence of local NFX they work as a sequential list of asymptotic network effects that may become saturated independently. When this is the case, the winner-takes-it-all rule plays in a local basis instead of globally and defensibility is local. Uber and most of the on-demand start-ups are very good examples of this.
1.4.2 Global NFX: in this case, a user joining in New York may increase the value of the marketplace for users joining in the opposite part of the world. My favourite here is Airbnb. As every host can be a guest and vice versa + every user can travel to any place in the world, the NFX are global and very strong.
There are many other classifications of NFX, as by the nature of the source: data, tech, etc but the ones above are a good sample to have a wide and clear perspective of the different nature of them.
Ideally, as a marketplace operator you should try to find ways to trigger different NFX and make the most out of them. Bear in mind that one business can present a combination of the different NFX mentioned above.
One of the beauties of NFX is that they can be extremely powerful and many times go unnoticed given their complexity.
2. Brand equity
A strong brand represents an intangible asset that can be an extremely powerful moat. According to Millwardbrown the top 100 brands account for 3.3 Trillion in value and growing above the market.
In marketplaces, the brand has to be built for the ’n’ sides of the marketplace. Many times we see marketplace founders forgetting about the fact that all sides of the marketplace have to be treated as customers, not just the user side…
Here is an interesting example of uber taking care of the supply side…
Sometimes it is just as simple as a two-sided platform (Ebay) but sometimes it can be a bit more complex:
Ubereats: a) Restaurants (supply-side #1) b) Couriers (supply-side #2) c) Buyer ( demand-side)
21buttons: a) Influencer (supply-side #1) b) Brands (supply-side #2) c) Buyer (demand-side)
When it comes to branding, another important point to take into consideration in marketplaces, is the classic brand positioning vs curation balance — as you may not manage the end to end process but the experience of the users in your platform will affect your brand positioning. That’s one of the reasons why curation and account management are usually key activities in the operation of a marketplace.
3. High switching costs
Another defensibility comes from switching costs, that is, the cost a user suffers when changing from your platform to another one.
There are different drivers, we identify 2 main groups:
3.1 Self or auto-personalisation: When the experience of a given user improves when using the product again and again. e.g. In Amazon, based on how much you use the platform, it recommends more and more precisely, and even allows you to use one click payment, etc. Coming back to food delivery, once you’ve had a few orders from 4–5 different restaurants, it is more likely that you repeat an order than building one from scratch; therefore, the friction of the process is reduced.
3.2 Embedding into the process of the users: SAAS-enabled marketplaces are a good example in this case. A piece of software that connects the demand or the supply side of the marketplace. That tool itself helps the user with one or more processes or to integrate with other processes. e.g. spotahome a home rental marketplace provides landlords with a SAAS tool to manage properties, and that’s a powerful tool to lock-up the supply side of the marketplace. Another example is Ontruck and the tools they provide to the demand-side (shippers) of the platform to track the shipments. Again one of my favourites.
At the end of the day, the aim of those two categories are commonly known as to “lock in” demand or supply-side users.
After speaking for a few minutes..
David asked: so… “would you invest in absence of network effects?”
I answered: yes, we would invest if other powerful moats do exist.
Graphically, I would go for the green triangle or the orange one but I wouldn’t go for the red one…
