This blog post is on what I call the “free parking” business model, which in my view is a different way of describing the “attach” business model. @cdixon tweeted a suggestion that I follow up our exchange of tweets with a longer blog post.
The right way to look at almost any business question is to start with the customer. Unfortunately, sorting out who the customer is may not always be simple to determine, especially in “platform” businesses. A platform is a type of market with multiple “sides.” In other words, a market maker that owns a platform acts as an intermediary between different “sides” of a market. While many multi-sided markets involve only two sides, others can involve several more.
The most powerful variant of a multi-sided market occurs when “sides” of the market are “interlinked” (i.e., attached) in a way which enables the harvesting of indirect “network effects” on one or more sides (technically known as internalizing positive externalities). Network effects, which are also known as “demand-side economies of scale,” result when a product or service becomes more valuable the more users use it. As an example, Google’s advertising serving business has network effects that are described by Warren Buffett and Charlie Munger as the strongest moat they have ever seen. Those networks effects can form the basis of a powerful “moat” which bestows sustainable competitive advantage for the owner of the platform such as Google. While a company may have revenue without a some form of a moat, it is highly unlikely it will be sustainably profitable.
Some network effects are strong and some are weak. For example, Google’s moat in the advertising-serving business is very strong. Yahoo’s moat in the financial news market (Yahoo Finance) is very weak. Some markets impacted by multi-sided makers with network effects are big and lucrative (e.g., Bloomberg’s terminal business; ESPN) and some are not (Yahoo Sports).
While some are quick to talk about network effects of the platform business, there is a great deal of context that determines the strength and sustainability of a most built on a platform + network. Elements of this context include ease of entry for all or part of the network, suitable, if different, replacements for elements of the network, and approaches by competitors that commoditize part of your system. That is why one must consider the full ecosystem and competitive dynamics around one company’s collected offering.
The existence of an “interlink” between sides of the multi-sided market is critical. For Sun, Java has network effects for users on one side of the market but there is no interlink to anything profitable. If anyone can take advantage of the network effects on one side of a market there is no moat benefit for the company, in this example Sun. In other words, Sun had no interlink at all to Java and so created a strategy that was fully suicidal. The Java strategy employed by Sun was like owning a yacht and casting off the lines and letting it drift down the river by itself. Red Hat makes a market by working hard on its “free” Linux distros so it can sell services, but anyone can do so which limits the power of the interlink.
One of the greatest challenges in creating a platform business is the so-called “chicken and egg” problem (i.e., one side of the market may not want to be participant in the platform until the other side is participating).
Experience has shown that the best way to accomplish the bootstrap of the “chicken and egg problem” is to give away one side of the market for free. This is particularly true in the software business since the marginal cost of offering up another “free” unit to end users is often close to zero. The free offering, if offered in the right way at the right time, can cause adoption that is both huge and swift creating network effects that can “tip “the free service in a way that creates a moat and barriers to entry (e.g., Facebook’s social graph). It is critical that the business not monetize too early like MySpace did before the market tips or all can be literally lost. This is where my “free parking moniker” comes into play. What is “free” is interlinked to what is profitable via an interlink which is like validated “free parking.”
It is important to consider that these free services (and this is really software—software eating the world, as Marc Andreessen has said) can be coupled with physical goods or businesses unrelated to pure play software/services. Fitness bands will offer more and more related software, but someone making a profit from supplements might choose to give away the band and the software if the margins on supplements are higher.
Experience has shown that if a potential market maker tries to charge a fee to both sides of a multi-sided market, especially in the beginning, they can’t get past the bootstrap of the “chicken and egg problem.” Typically it is best to offer the free side to consumers since no one loves “free” more than a consumer (for a number of reasons described best by behavioral economics). It is also best to give away as free the service that has the lowest marginal cost. In many cases web services have almost no marginal cost after they are developed and can an ideal “free” offering. .
One consequence of the “free parking” business model is that you run a risk that some company deciding to give away what you sell, especially in a technology business where there can be zero marginal costs of giving away a service. Free services are being used in ways never imagined before to make interlinked services more profitable. To use the most recent version of Android, you are locked into Google’s advertising serving service and all of the other Google services that feed into that advertising service. Google is interesting in that instead of having multiple services as profits centers, all of the services are in effect a web services tornado sucking in everything it can reach to benefit serving advertising.
Freemium is a variant of the “free parking” business model but there are several sub-variants of freemium itself. In many cases freemium is used since the company has relatively few dollars for marketing and sales. The company in this situation transforms code into the equivalent of marketing spending and “gives away for free” service X to generate qualified leads for interlinked service Y. Freemium works best if service X has network effects. But not all freemium services have strong network effects in a big lucrative market. Giving away salty peanuts in a bar to sell beer as a loss leader is not a case involving network effects. By contrast, giving away a Instagram to end users did create network effects. Instagram is interesting in that it never actually created in interlink to another profitable service before the company was sold.
Many examples of freemium exist as services that have some free level and then a paid-for level that is implemented as a subscription or time-based payment. Physical goods don’t make particularly good free products and the strategy of free entry level products with a one-time transaction for a purchased product are painfully difficult product lines to manage (especially in software). Freemium, as attractive as it is, can be a higher risk strategy down the road. Businesses depending on freemium fight a battle on two fronts, month-by-month. They must spend energy on a broad “air attack” to bring new customers in, presumably at the free level. They must also fight a ground war which is to keep existing customers connected at a paying level.
In order to make these battles easier to fight it is highly desirable if not essential that the free offering help create some sort of a moat even if network effects can’t be created or if the network effects are weak. Costco is an example of a company which sells goods at close to break-even financially so that it can sell the memberships that bring in 80% of its gross profit. Costco has created “supply-side economies of scale” that are tough to match. While giving away goods and services is easy to do, creating a moat while doing so is a genuinely hard problem. For a blog post in which I have discussed other sources for a moat see: http://25iq.com/2012/12/06/charlie-munger-on-moats-first-of-the-four-essential-filters/