What Do We Know About Unicorns? Defining and Finding Unicorns

This is a companion article to my LinkedIn posts on #Unicorns. You can find the first post here. I provide here more background and details behind the results in the posts. I start by explaining how I constructed the list of #venturecapital backed unicorns. In the next article I will describe the definitions and measurements of the unicorn geography. You can find the full-length pdf document here.

Note: I often use “we” in the text: this is not intended as a regal “We,” but the acknowledgement of the fact that it was and continues to be a team effort. We started tracking unicorns at the Venture Capital Initiative at #Stanford back in 2015. This is an ongoing project and I will be expanding and updating it frequently. Stay tuned.

1. Defining a Unicorn

To analyze unicorns, we need to find them and construct a comprehensive sample of them, and to do so we need to define a unicorn. Defining a unicorn in an unambiguous way is more difficult that it may seem. There are many available lists of unicorns [1], and even a cursory look would suggest that there are a lot of companies on some lists but not others, and that some lists are much longer than others. However, without a clear definition, it is challenging to accomplish the first goal of the scientifically based analysis: construct a unicorn sample without obvious omissions and (explicit or implicit) selection bias. It is important to note that while no definition is perfect, clarity has at least two important virtues. First, the sample is verifiable (that is, it is possible to check if the company satisfies the chosen definition or if a company is missing). Second, if the definition is clear, it is easier to understand and discuss the implications of any analysis.

Therefore, I first define a unicorn and identify specific inclusion and exclusion criteria. In coming up with a unicorn definition, I have to trade off simplicity and verifiability, on the one hand, and some economic rationale, on the other hand. The sample inclusion criteria are clear and the sample can be replicated and verified by independent analysis. To this end, I publish the complete lists of unicorns that satisfy my criteria [2]. Then, armed with these criteria, I proceed by constructing a comprehensive sample of unicorns that satisfy my definition.

I work with two definitions of unicorns and this gives rise to two main samples, which I call the USMAIN sample and the USPRIVATE sample. A company is a unicorn in the USMAIN sample if it satisfies the following three conditions.

Condition 1: VC funding. The company must have at least one documented venture capital funding round since January 1, 1995. The reason is simple: we are interested in high-growth innovative companies that have raised venture capital funding. Investments by both institutional and corporate venture investors qualify. This is not a very stringent condition and most potential candidates qualify. An example of a company that is included in some unicorn lists, but does not satisfy this criterion is Supreme. Based on publicly available information, Supreme never raised a VC funding round.

As we know way less about companies from decades ago, 1995 is a calendar threshold based on data availability. Available evidence suggests that most candidates are young enough and 1995 is not an overly stringent condition either.

Condition 2: U.S. location The company must have been headquartered or legally incorporated in the United States at the time it became a unicorn. There are obviously many unicorns elsewhere, such as in China, and so this is entirely a data availability criterion. While it would arguably be possible to create a comprehensive sample of non-U.S. unicorns, I am interested in analyzing many interesting dimensions of unicorns and their founders, and collecting data on them would require more effort.

This condition is thus a temporary condition that I hope to relax sooner rather than later to construct the GLOBALUNICORN sample. If companies were founded elsewhere but later relocated to the United States, such companies would be included in the current sample if the relocation took place prior to or at the time the company became a unicorn. For example, Slack Technologies was founded in Vancouver, Canada, in 2009. When the company became a unicorn in 2014, it was headquartered in San Francisco, CA, and is thus included in the list.

Condition 3: Valuation. The company must have had at least one private funding round with a post-money #valuation equal to or above $1 billion or should achieve at least $1 billion in valuation at the time of company’s first liquidity event, in both cases prior to January 1, 2020.

“Private” means that at the time the company raised a funding round, it was a privately owned company. A “liquidity” event means either a public listing (IPO, direct listing, or reverse merger/SPAC) or a sale. It is important to note that the company should achieve at least $1 billion valuation at its first liquidity event. For example, if a company becomes public at a valuation below $1 billion and later is acquired for more than $1 billion, this company is not defined as a unicorn. There are a number of reasons for such a restriction, one being that the ownership structure of such a company is often vastly different compared to the time of its first liquidity event. An example of a company that does not qualify is Amazon. Amazon raised its VC round in 1996 and went public (that is, experienced its first liquidity event) in 1997 at a value of about $330 million. Of course, no definition is perfect, but as I said above, clarity is a virtue.

The cutoff date (currently January 1st, 2020) is a moving target. The sample will be updated and expanded from time to time. Obviously, many unicorns should be added for 2020 and 2021.

The definition of post-money valuation is the product of the share price in the most recent funding round and the total number of common shares on a fully diluted basis. I use post-money valuation, because this measure is more likely to have been reported and easier to verify, and it attracts a universal following in the start-up world. Will Gornall and I showed that the fair value of VC-backed companies tends to be below that of their post-money valuation [3]. As a result, many companies with post-money valuations of $1 billion and over would be fairly valued at well below $1 billion, and thus would not qualify as unicorns if the fair value methodology was used. Pre-money valuation in many cases would be a better measure for consistency, but data availability is tougher. Of course, any specific value used in the sample construction is an ad hoc threshold. Post-money valuation has an advantage because of its simplicity and wider availability.

To qualify for the inclusion, it should be possible to ascertain that the company’s post-money valuation is $1 billion or above with a high degree of confidence. An example with an uncertain valuation outcome is Trusteer, which IBM acquired in August 2013. Pitchbook reports an $850 million post- money valuation, while VentureSource reports a $1 billion post-money valuation. TechCrunch stated that it was a $1 billion deal according to sources close to the deal. The Israeli financial paper Calcalist reported the deal in the range from $800 million to $1 billion. Reuters announced that IBM was paying close to $1 billion. As a further investigation could not establish that Trusteer had a liquidity event at $1 billion or above and it did not have prior unicorn rounds, the company is not included in the sample [4].

If the company satisfies all the criteria above, it is a unicorn in the USMAIN sample. A company is a unicorn in the USPRIVATE sample if it satisfies Condition 1 and Condition 2, and if Condition 3 is tightened to Condition 3’:

Condition 3’: Valuation. The company must have at least one private funding round with a post-money valuation equal to or above $1 billion prior to January 1, 2020.

“Private” excludes companies that became billion-dollar companies only at the time they became publicly listed or at the time when they were acquired. This is why this sample is called USPRIVATE. Note that these companies could now be public or private, or they may no longer exist. Note that all the companies that are included in the USPRIVATE sample are also included automatically in the USMAIN sample.

For example, Tesla achieved a $1.31 billion pre-money valuation at the time of its IPO, but its most valuable private round valued it below $1 billion. Thus, Tesla is in the USMAIN sample, but not in the USPRIVATE sample. As another example, Oculus (now Facebook Technologies) was acquired for around $2 billion, but its highest valuation in private rounds was below $300 million.

2.  Finding Unicorns and Unicorn-Related Data

To find all the unicorns, I combined all the available commercial datasets, such as PitchBook, Venture Source, Thomson Reuters VentureXpert, and Genesis, as well as several lists of unicorns in various media publications. My team and I also acquired and painstakingly coded their numerous corporate charters. For each candidate, each of the three conditions was carefully checked and cross-checked [5].

The final USMAIN sample contains 531 unicorns and the final USPRIVATE sample contains 354 unicorns. These samples include household names such as Facebook and Uber. It also includes perhaps less familiar names such as WebMethods, Tellium, and AirWatch. (Have you heard about the last three? I had not before I started collecting the unicorn data.)

 For each unicorn and related variables (such as the unicorn’s founders and investors), my team and I manually collected, constructed, and cleaned at least 150 variables (and yes, I plan to share the results on many of them over time).

How do these samples compare with other unicorn lists? CB Insights, for example, has been providing unicorn lists every year since 2015. CB Insights lists only unicorns that are private at the time of reporting, making it comparable to the USPRIVATE sample. The first recorded unicorn event in the CB Insights list took place in 2009, compared with 1999 in the USPRIVATE sample. Combining its samples for the years prior to 2020, the merged CB Insights list contains 302 unicorns, compared to 354 unicorns in the USPRIVATE sample. In some cases CB Insights includes companies, for which the unicorn valuation could not be confirmed [6]. For example, CB Insights included Collective Health with a date of joining the unicorn club of June 2019, but the first confirmed deal with a $1 billion valuation took place in 2021. Another example is Flipboard. The company raised its Series D in July 2015. TechCrunch reported that “Flipboard did not provide valuation for this round; according to the filing, the valuation could range from $800 million to $1.32 billion.” CB Insights included Flipboard with a date of becoming a unicorn of July 2015 and a $1.32 billion valuation. Pitchbook shows a $900 million post-money valuation in this round. As it was not possible to confirm the precise valuation, Flipboard is not included in my sample. In some other cases, I could not confirm the valuation of the company at all. For example, CB Insights reported Zhangmen as a unicorn in 2017, but this information could not be independently verified. For around a dozen unicorns VC funding rounds could not be confirmed either.

Recently, Ali Tamaseb of DataCollective published a thoughtful analysis of 195 unicorns and their founders using a dataset of 65 factors he manually constructed (and given all the efforts we have gone through this is truly a feat!). His definition is somewhat different [7]. He selects U.S.-originated companies founded between 2005 and 2018 that at one point (may have) passed $1 billion in valuation. His sample is thus not just limited to unicorns (in my definition) but also includes any #startup that ever passed $1 billion in valuation. For example, Nevro which received a $1 billion valuation in the secondary public offering is included in Ali’s sample, but as it was valued only at $306m at its IPO, it is not included in mine. For some of the companies in his list, the unicorn valuation could not be confirmed and thus they are not included in mine. For example, the pre-money valuation of HubSpot was $634 million on its IPO day on October 9, 2014 and the company reached its maximum post-money valuation of $914 million on that day [8]. Tamaseb also included some #startups that, according to my evidence, have not raised VC funding, such as Acadia Healthcare and Carvana. However, there is obviously an overlap in some of the questions we look at and I encourage everybody to look at Tamaseb’s discussion as well.

In the next article I will describe the definitions and measurements of the unicorn geography.

Notes

Many people have participated in making this project a reality. I would like to thank especially Alexey Dang, Will Gornall, Anastasia Sochenko, and Amanda Wang for many useful suggestions and all their help. Anna Elyasova, Michelle Gan, Jeremy Guenette, Matthew Hamilton, Daniel Ibeling, Simrin Kalkat, Kendall Kissell, Kevin Lee, Ian Lim, Alex Lukashin, Dev Narang, Sarah Probst, Victoria Valverde, and Georgiy Zvonka provided valuable research assistance. I thank the Venture Capital Initiative at the Graduate School of Business, Stanford University for financial support. Please address comments to istrebulaev@stanford.edu.

[1] See, for example: The Complete List of Unicorn Companies, CB Insights, 2021 or The Unicorn Club, Dealroom.co, 2021.

[2] As I discuss later on, in some cases the inclusion decision was not clear cut. As I am alerted to new evidence, I will update and revise the unicorn list. In particular, if you possess verifiable evidence that a company on the list should be excluded or a company should be added to the list, let me know.

[3] Will Gornall and Ilya A Strebulaev, “Squaring Venture Capital Valuations with Reality”, Journal of Financial Economics 2020, 135, no. 1, 120–143 [Available online]. A follow-on paper contains an expanded version of the valuation framework.

[4] Techcrunch article, Calcalist article, Reuter’s announcement.

[5] There are quite a few unclear cases, often related to either inconsistency among our data sources that was impossible to resolve completely or the backfilling, practiced by some data sources. For example, in April 2019, Pitchbook reported that Sila Nanotechnologies achieved a unicorn status in its completed Series E funding round. CB Insights also reported that the company became a unicorn in April 2019. I included Sila Nanotechnologies in the USMAIN sample. Pitchbook later changed Sila’s post-money valuation in that round to the lower value of $938 million. Sila raised further rounds since then, including a unicorn-level round in 2021.

[6] In its 2017-2019 lists, CB Insights mention that its lists include companies with “whisper valuations.” I have no doubt that some of these whispers are correct, but it would be an overreach to use them as a basis for constructing my list.

[7] See his article here. Recently, he followed up with a book: Ali Tamaseb, Super Founders: What Data Reveals About Billion-Dollar Startups, 2021.

[8] See Business Insider article and Techcrunch article.

All http links were accessed and checked on September 21, 2021.