The Mechanics of Unbundling Part II - Platform Dimensionality
A Deep Dive Into how Horizontal Platforms Truly Become Unbundled.
If you’re reading Bits & Bytes for the first time, I recommend starting with Part I. If you’re a recurring reader: Welcome back! Special thanks to all who reached out and asked questions. My hope is that this content sparks thoughtful discussion.
Here’s where we’re at in our journey:
The Mechanics of Unbundling:
Part II - Platform Dimensionality
Part III - Network Effect Types & Strength
Part IV - Putting the Pieces Together
In Part I, we built the foundation for understanding whether or not a vertical solution could disrupt a bundled incumbent. We examined an individual subnetwork and established that carrying capacity, or the threshold that a platform can effectively support suppliers and consumers to allow them to transact effectively is essential in determining if an isolated category is underserved. This underserved category provides the seed for a vertical solution to grow a vertical marketplace.
Monolithic platforms aren’t a singular network but a network of subnetworks. To deepen our understanding of unbundling, our analysis must extend beyond the myopic view of an individual sub-network to encompass the entirety of a platform. In addition, incumbents are also dynamic and can respond to vertical threats. Therefore, our analysis must also help predict whether or not an attack from a vertical challenger will likely elicit a response from the incumbent.
Our analysis finds that a spectrum of vulnerability for incumbents exists. It depends on the number and variance in its carrying capacities. The more carrying capacities (i.e., subnetworks) that exist and the more heterogeneous they are, the more likely the incumbent will face vertical challengers. This spectrum is also proportionate to understanding whether an incumbent is expected to respond with a countermeasure. We will explore these mechanics in greater detail.
Quantity of Carrying Capacities
As previously noted, when an isolated vertical’s carrying capacity has been eclipsed, the sub-network becomes saturated (for those that didn’t read part 1 - the platform cannot adequately support the quantity of supply and demand), and an opportunity exists for a challenger to build a vertical marketplace. However, the amount and heterogeneity of carrying capacities make the incumbent’s bundled platform slow to react and indicate how ripe the opportunity is to unbundle. The more carrying capacities (i.e., sub-networks) a platform has, the more fragmented it is. The more fragmented the platform is, the more challenging it is for the incumbent to cater to each sub-network as it grows.
The number of carrying capacities present on a platform is measured by answering two questions:
How many use cases or types of supply do the incumbent platform support, and how heterogeneous are they from one another?
What type of network effect does the platform possess: Local or global? In other words, is the platform only valuable for supply and demand localized to a specific geographical area?
Answering these questions for any platform provides insight into how defensible it is and how attractive it is for a challenger to take a vertical strategy to disrupt it. The former question we will discuss in Part II, and we will address the latter in Part III.
Platform Dimensionality
Rarely are platforms heterogeneous by design from their origination. Craigslist and eBay are the anomalies, not the norm. The initial breadth of each platform can be explained by the lack of competition at the time of its founding. In the mid-1990s, there weren’t options for buying and selling in your community (Craigslist) or transacting collectibles and other goods online (eBay).
More often than not, heterogeneous platforms are a product of the natural expansion of the initial network. As the network expands, niche networks and unexpected use cases form. Alternatively, as a platform begins to hit market saturation, a platform will layer on additional products and services to meet increased demand and continue growth in a market. Each is an example of bundling - the diversification and expansion of suppliers and use cases to maximize user engagement and lifetime value (LTV).
Each use case or type of supply or use case can be considered a dimension of a platform.
A dimension exists when the type of supply or use case would require specific and unique product enhancements to increase the carrying capacity (the threshold that a platform can support that allows suppliers and buyers to transact effectively).
Example of platform dimensions:
eBay: Used clothes, new clothes, furniture, electronics, etc.
Craigslist: Legal services, appliances, vacation rentals, education jobs, etc.
Airbnb: Vacation rentals, experiences
As we’ll see, the more heterogeneous dimensions a platform possesses (i.e., the less overlap between them), the more vulnerable it is to being unbundled. Conversely, the more dimensions a platform has, and the more homogeneous they are (or, the tighter the overlap between them), the more challenging it will be to disrupt. Platforms that possess dimensions with low overlap between their dimensions we will consider highly dimensional, and platforms that have several dimensions but high overlap we will consider as low dimensional.
As the breadth of the highly dimensional platform expands, it becomes increasingly challenging for the platform to cater to each one as they will each have distinct needs and challenges to increase their carrying capacities. As a result, product enhancements are often released to cater to the broadest set of users or for the most significant number of dimensions the platform contains.
Let’s examine a theoretical example.
Let’s say we have two platforms, one horizontal (high dimensional) and one vertical (low dimensional), thinking about where to invest engineering and product resources. The horizontal platform allows suppliers to resell secondhand clothes, furniture, shoes, and re-used vehicles. In addition, it will enable manufacturers to sell new goods such as electronics and sporting goods. Each dimension represents an equal share of the revenue for the platform. The vertical challenger is only focused on selling re-used vehicles.
Each platform will invest where it can gain the most market share. To achieve the most market share, it makes sense to concentrate capital on investments that will maximize the network's overall carrying capacity, allowing them to capture more of the addressable market.
The horizontal network will look to maximize the coverage of the dimensions affected. The more dimensions affected, the higher the return on invested capital. After weighing the options, they settle on a product enhancement that can increase the carrying capacity of 3 dimensions at once, tripling the ROI on the investment. It’s a tool that automatically removes the background from the good that a supplier is posting for sale - increasing the attractiveness of the listing and the likelihood that a consumer makes a purchase.
This product enhancement could benefit the carrying capacity of dimensions such as clothes, shoes, furniture, etc. That said, it may not necessarily help suppliers selling used vehicles, new electronics, etc. By investing in a broader number of dimensions, the high-dimensional platform forgoes investing in a dimension requiring more specificity, such as used vehicles. Other examples could include the authentication of a ticket, the scoring of how good of a deal it is for purchasing tickets, or the scoring of the cut and clarity of a diamond for jewelry.
In the case of the vertical challenger, all resources are dedicated to increasing the carrying capacity of the sole network it supports: Used Vehicles. Therefore, the vertical challenger will choose to invest in the product enhancement that increases the carrying capacity of the network the most. In this case, the enhancement could be providing a delivery service for the transacted vehicle.
This figure below is a depiction of the hypothetical example that plays out.
Note that attainable market size is critical in determining where investment resources will be allocated for the incumbent.
We will put aside the theoretical examples and explore this in reality, but before continuing, you’re probably wondering, at this point, how do I assess the heterogeneity of platform dimensions?
Although this may not be a perfect answer, it provides a proxy for how I think about it. Products & use cases require a range of potential product enhancements of varying specificity to increase their carrying capacities. Think about how the experience of booking an Airbnb would be incomplete without reviews, seeing the available check in/check out date, or seeing high quality photos of the space beforehand. Or, the peace of mind by shopping through StockX because each product is authenticated, the discovery because each SKU is standardized, etc. Each is an example of specific product functionality to increase the carrying capacity of the dimension(s) on the platform.
The more specific the enhancements are to that particular dimension across the five characteristics, the more likely it is to be heterogeneous when inserted into a pool of other dimensions. The emphasis either side of the platform puts on one of these characteristics influences how quickly a vertical could be splintered off.
Authenticity/Quality - Is there skepticism that the product could be fake or not to the buyer's expected quality? Examples could be phony inventory, poor work quality in a purchased professional service, etc. Building features to establish consumer trust is often no easy feat.
Price - Is the price dynamic? Does it depend on other suppliers? Appropriate pricing is crucial for suppliers to quickly maximize their income and turn inventory. Examples include cars, tickets, and homes.
Fragmentation & The uniqueness of Supply - How many unique units of supply are available? Unique supply is often challenging to acquire but important to consumers.
Discoverability/Time to value - How challenging is it to build features that allow users to find what they’re looking for? Examples include unique taxonomies/or granular search, date ranges, and maps in the case of looking for rentals, qualifications for professional services, etc.
Alternative Services - Are other services required to complete the transaction? Examples could be delivery, financing the purchase, or human intervention.
Commodity goods, such as electronics, books, home goods, etc., are generally less specific. The lack of specificity is because prices are primarily static, authenticity isn’t often a question, and there aren’t constraints on the abundance of supply. It’s also generally easier to direct consumers to what they want.
High-intent purchases are higher in specificity. Homes, cars, and tickets are dynamic in price. They depend on many factors, and suppliers want to optimize their gains. So it’s easy to compare and contrast.
Ultimately, this means that when a platform is servicing dimensions with various specificities, it will struggle to optimize the customer experience because it will be forced to choose where to invest its resources. In addition, investing in the required enhancements might also introduce asymmetry into the product and make it disjointed.
eBay versus Stubhub exemplifies how challenging it can be for an incumbent to react and compete on a dimension with high specificity.
For tickets, there’s a range of product enhancements to increase the carrying capacity, each non-trivial to implement.
Authenticity/Quality - Consumers want to ensure that the ticket is authentic.
Price - Tickets have dynamic pricing, which depends on many factors: supply available, price of supply available, and the amount of time before the event. This dynamism is essential to account for in optimizing the sale for suppliers.
Uniqueness of Supply - Each unit of supply is unique and provides a different consumer experience.
Discoverability - Maps of the stadium, previews of the seats, filtering by the best available deal, etc. Each influences whether or not the consumer makes a purchase.
If you’re eBay, investing significant capital in these product enhancements is challenging, given that it affects a subset of your overall network. However, if the dimension represents a significant revenue share of the platform, it’s easier to justify an investment. Frequently, dimensions make up a fraction of revenue, as was the case for eBay and tickets. Chances are, investment dollars are better allocated to product features that affect multiple dimensions and produce what seems to be, at the time, higher ROI. Or, the decision is made to acquire the challenger, as was the case for StubHub in 2007 for $310M.
The smaller the revenue share of the dimension, the less overlap with other dimensions, and the more complex the product enhancements are, the less likely the incumbent is to invest in the countermeasures to combat a vertical challenger.
Over time, we’ve seen eBay retreat from dimensions with high specificity and opt to invest in dimensions with low specificity and high overlap, thereby strengthening the value of their bundled offering. eBay’s revenue is comprised almost entirely of commodity goods. Shoes, Books, Home & Kitchen, Tools, etc., all dimensions with low specificity.
*Note that automotive is an exception here. First, it’s worth noting that eBay sells more parts than cars. But, because automotive is so large, eBay has invested significantly in the product enhancements needed to maintain a high carrying capacity (such as financing options, logistics, discovery, etc.)
Analyzing Craigslist's high dimensionality also explains why we haven’t seen them react to being unbundled. Each dimension on the platform is highly specific. The product enhancements to increase the carrying capacity for legal services are disjointed from the enhancements required to increase the carrying capacity for vacation rentals. So on and so forth for selling goods, finding a parking space, and discussion forums.
When a vertical challenger, such as Airbnb, decides to compete on vacation rentals, it’s unlikely for Craigslist to devote special attention to the dimension, given that it makes up a fraction of the overall listings on the platform.
Platform dimensionality also explains why we see a few big winners in Craigslist's unbundling instead of several equal-size winners per category. Due to low variance in dimensionality in each sub-category, companies such as Thumbtack, Zillow, Reddit, OfferUp, etc., have horizontally integrated an entire category.
Platform dimensionality alone doesn’t give us the complete picture, though. Are Uber and Airbnb equivalent despite having low dimensionality? Is Craigslist so similar to eBay? No. They’re different because each network has distinct characteristics. Each platform possesses a different network effect that affects how fragmented it is.
We will describe how the type of network effect affects the likeliness of a platform being unbundled in Part III.