Screening 142,731 Companies

When I joined Hummingbird, I decided to build a 'startup radar' which will help us screen and source companies on a global scale. Now, some of you might be asking — why wouldn't you just use one of the already available solutions? I have written a bit about that in my previous post, but in brief; I simply don't believe that you can achieve extraordinary results if you take the same steps as everyone else.

Moreover, I think the best way to start as a VC is simply to go through tens of thousands of companies to develop; (a) a better understanding of the ecosystem, (b) your own thesis about different verticals, and (c) a touch for outstanding companies. All of those, of course take time and commitment, but in the past I have learned that the most important learnings come from sh**load of data.

So what have I learned?

1. Mining data is pretty much the same as coal mining

You need to know where to dig, and you need to be prepared to dig through a lot of soil before you see any sort of results.

Knowing where to mine is in my opinion like 70% of success. Knowing how is like 25% and the reaming 5% is the actual screening / processing of data.

There are no shortcuts to this, you will simply need to test a bunch of sources before you develop a high quality dataset.

However, there are two fundamental differences between data and coal mining. And that is; (a) data is typically scattered among a lot of different sources. Thus, your actual dataset won't really come from one “coal mine”, but rather like 15 different sources and APIs. (b) Company data is a living 'organism'. Companies grow, evolve, pivot or die. So data mining is and endless, ongoing process.

2. Signal vs. Noise Ratio is higher than ever

To put things into perspective; from this first (somewhat curated) dataset of 142,731 companies only 480 passed my initial screening. That is roughly 1/300 companies. And from those already over 360 didn't pass my second screening (meeting/call). I think realistically I am going to end up with ~20 interesting companies. To put things into perspective again, that is 0.014% or roughly 1 in 7500 companies.

Now, this is of course a completely subjective ratio which depends on many variables — like traction, funding stage, or industry focus, but in general I believe nowadays VCs have to go through more noise to find signal than ever before.

There are many underlying reasons for this, but perhaps the chart below might give you a hint…

3. Variables, variables, variables

As one of my colleagues, Barend, says; there is no magic formula to venture capital. And I have to admit, that scares the hell out of me. You see, when I joined Hummingbird, I naively thought I'm going to mine bazillion companies, figure out this special secret formula to score companies. And then I'm just going to sit back and let the machine find the next Facebook for me. Truth be told, somewhere back in my mind, I still believe it's possible. However, I no longer believe the solution is a simple formula or an algorithm, but rather a data collection process that happens in phases.

One thing I learned about company variables is that they really do change a lot. And those changes provide a lot of context — which can be actually a hell of a lot more valuable than the variable itself.

Secondly, I learned that you need to be absolutely brutal about decisive variables. If you can't collect a certain variable in an accurate manner, just drop it.

And third, look for patterns from the very beginning, but try to avoid gambler's fallacy as much as you can.

4. The myth of the last unattended vertical5. “There are no competitors to what we are building”

Whenever I hear someone saying that nobody is doing what s/he is doing, I imagine this cute unicorn hitting a wall at full speed. Because that's what will happen to this “no-competitor belief” once you actually conduct a thorough research. Only in US, there are approximately 30 million, I repeat 30 MILLION businesses. If you honestly believe you have no competitors, shoot me an email — I will be happy to share a few with you.

And as much as I don't believe in “no-competitors”, I also don't believe in “last unattended verticals”. There is no such thing as last. As a society we evolve, technology evolves at a rapid pace, and our needs completely change. I don't even dare to guess how is the society going to look like in 5 years. So “last” is just something people say to create hype and sense of urgency.

I hope you enjoyed this article. If you'd like to know more about my radar, or if you'd like to get a sneak peek at some of the fastest growing companies, sign-up for my newsletter or drop me an email ;)

Written by Dominik Vacikar.

Views are my own and don’t reflect the views of my employer.

Feel free to contact me at;[email protected]