In this part, I’m going to combine some rules of thumb to predict the success of a startup. The model will be based on three of the 
four characteristics that Peter Thiel feels are important for a startup to achieve a monopoly (take a look at chapter 5). The model may give some insights into why it might be that venture capital has tended to be focused on internet, biotech and communications as opposed to cleantech, industrial or consumer goods.
My idea of a startup* is to sink in some money early on in order to reap a massive reward later – aka massive profits. In a perfectly competitive market there would be no massive profits, so, as a startup you have to find a way to be in an uncompetitive market – aka a sustainable monopoly. Thiel has come up with what, at least to my naive self, seem to be four good fundamental characteristics that lend themselves to ensuring a position of monopoly. Therefore, I’m going to take three of these four characteristics, define them on a scale of zero to one (no pun intended) and then combine them into a single predictor of startup success (scored out of three).
1. Intellectual Property
The idea here is that if you have a patent or a trade-secret, you, by definition, have something that no one else has and that you can charge a premium for. At the high end of the scale I would say are companies with chemistry or pharma based patents or trade secrets – ones that are either very hard to copy or else are really obvious if someone copies the patent so you can sue them; such startups I would score 1/1. In the middle I would put patents that are easy to copy but somewhat possible to track if other people copy (0.5/1). At the bottom (0/0) I would put software based companies whose functionality could be built from scratch if you had enough dollars.
2. Economies of Scale
If the marginal cost of what you make falls as you make more then you are in a position to undercut competitors who are smaller in size. This can prevent competition. At the top end of scale (1/1) I would put software, which has a marginal cost of about zero. In the middle (0.5/1) I would put hardware or infrastructure, which doesn’t have the economies of software but still generally falls in cost with volume. Then, at the bottom (0/0), I’d put a service company, whose costs basically scale linearly with volume. [I’d also consider information databases to be part of the economy of scale of software, although I could understand arguments that they should be considered intellectual property].
3. Network Effects
Facebook and Uber and eBay are all really useful because there are lots of users. The more users the more useful the startup is (I’ll talk about Metcalfe’s law in a later blog). Once lots of people sign up it doesn’t make much sense to switch to a competitor and this has the effect of warding off the competition. The scoring on network effects is fairly simple; 1/1 for any marketplace or communications company and 0/0 for all other companies.
There is a fourth characteristic that Thiel feels is important for monopoly – brand. I think (again, naively) that what Thiel says about brand enabling profits is good stuff (e.g. Coca Cola or Burger King). However, I won’t include brand here in my rules of thumb because I don’t think you can measure whether a startup will have a strong brand or not when it is at an early stage.
Now, let’s look at what this model says about different types of startup:
- Internet/communications: I’m lumping internet and communications into one because they all score 2/3 (or debatably 2.5/3 if you include strong algorithms as intellectual property).
- Biotech/Pharma: Most pharma startups will score 1.5/3 on this test. They have very strong intellectual property and some economy of scale but no real network effects.
- Consumer goods: Here I’d say 0.5/1 on IP (not as good as pharma), 0.5/1 on economy of scale and 0/0 on network effects, for a total of 1/3.
- Industrial goods: Here I’d say 0.5/1 on IP (perhaps 1/1 for some proprietary chemistries/materials), 0.5/1 on economy of scale and 0/0 on network effects, for a total of 1/3.
- Cleantech: Here it’d be 0.5/1 on IP, 0.5/1 on economy of scale and 0/0 on network effects for a total of 1/3.
The message here is pretty simple; internet, communications and biotech tick more of the boxes for achieving a monopoly position, and hence, I hypothesise, for success as a startup. As a consumer goods, industrial goods or cleantech company, there just aren’t the same number of avenues to monopoly.
Perhaps this model helps to explain why VCs have historically focused their dollars on internet, communications and pharma; or, perhaps you think this model is just Peter Thiel (with or without my further damage) looking at the data on startups and then backing out these characteristics. Either way, there are massive differences in the successes of different types of startups and I absolutely think it’s worth the effort trying to better understand why that is the case.
*to be clear, when I say my idea of a startup, what I really mean here is my idea of a startup as a (capitalist) tool to fulfill a mission.
~~~
In Part 1, I described how startups involve lots of diverse and incomplete information. For this reason, I argued that it’s important in startups to avoid opinions and gut feelings and focus on following simple rules of thumb (like I’ve tried to give in this part). I gave the example of how Ashenfelter, a wine enthusiast and statistician, by understanding the simple relationship between weather and wine prices, was better able to predict the future price of fine Bordeaux wines than wine tasting experts. Here I’ve included a key table from his 2008 paper.
From Ashenfelter’s 2008 paper. You can see here how the prices at which wines are traded very often converge towards the price Ashenfelter predicts using simple rules of thumb for age and weather during the year of harvest.