Will predictive software “eat”
the VC Manager?

What a difference 40 years makes? If you look at a variety of industries, you would see mind-boggling changes: just think telecommunications, computers and television.

Software appears to be “eating” everything. Well, everything except for the Venture Capital Industry.

Let’s go back 40 years ago. An entrepreneur pitching a VC didn’t use a PowerPoint in 1975, but he did whatever his entrepreneur grandson would do today: have his idea in some written form, and would try to get a meeting with a venture capitalist who might fund it.
That’s one thing when you have to handle, say, a 300 of investment ideas a year in 1975, but that’s a whole different ballgame when leading VCs are being approached 30,000 times per year in 2015.

Arlo Gilbert wrote an interesting post, arguing that “whether venture capital, accelerator, traditional angel money, or online angel money, the ability to get funded is only partly about how good your idea is, it is largely about who you know and how well you present. This is the problem.”
(https:[email protected]/silicon-valley-s-dirty-secret-67b1f0efdce)

Silicon Valley’s Dirty Secret

Gilbert claims, that the problem of investors (i.e. Silicon Valley’s Dirty Secret) is that they rely solely on human relationships to pre-screen and make investment decisions, and that arguably prevents phenomenal ventures such as AirBnB from getting funded by 7 leading VCs (Read Brian Chesky’s brief, yet powerful post https:[email protected]/7-rejections-7d894cbaa084).

While I agree with his analysis of the problem, I don’t consider his solution feasible (Gilbert calls for an automation of the venture capital investment process). It may seems extreme, because it is.

VC Partners are not the problem. They are usually very experienced, very smart inidivuals. But the challenge is that venture capitalists are incapable of serious considering or screening all of the proposal they receive, regrettably, even those they receive from people they know. There are just too many of them. So what can be done?

Examples of VCs foray into predictive analysis

Many leading VCs are using Data Scientists and Analysts, to try to beat their competitors to the punch. Kleiner Perkins Caufield & Byers (KPCB), one of the top VCs firms in the world, is using DRAGNET, a “social monitoring” system. This systems ingests data from like AppStore, Google Play, AngelList, Twitter mentions in order to hear about deals before it comes through the referral network.

Another VC who similarly uses social media parameters is eVentures.
Floodgates (Formerly known as “Maple”) is using a “Chief Hacker”, while Ironstone puts its faith in “growth science”.

More rigorously scientific approaches are taken by Correlation Ventures, Google Ventures, and Deep Knowledge, a Hong Kong based investment company. Correlation — The first pure “Quantitative VC” — and Google use predictive analytics to analyze data of successful start-ups. Deep Knowledge’s VITAL machine learning algorithm searches for patterns by comparing the parameters of the prospective investment (in life sciences) with available info relating to financing, clinical trials, intellectual property and previous rounds of funding.

The issues of existing predictive analysis attempts

There are several issues with these aforementioned approaches, but we’ll briefly note the main ones:

First, “big data” is a great buzzword, but no matter how good your predictive algorithm is, data without context is almost meaningless.
(For more discussion on context http://followtheseed.vc/2015/08/the-forgotten-question/).

Another issue is that of competitive advantage: you cannot understand if a venture has an “unfair advantage” by looking at aggregated data. Averages seldom tell you the story, even more so in the early days.

So, how do I suggest VC partners handle 35,000 meeting requests a year with entrepreneurs?

Before we answer that, here’s 2 questions to consider:

1. Have you ever asked yourself how come most successful and profitable brands, products and services are not the best or the cheapest (Think McDonald’s, Vodka Absolute or Starbucks)?

2. And how is it that companies that have products that are very easy to copycat(Think uber or Whatsapp) are worth 50 and 22 billion respectively?

What are Raving Fans?

Well, the answer probably lies in connection between the users and the product/service. More specifically, in people who irrationally love products and services, despite their being better alternatives available in the market.

They are not simply fans of those products and services: they are “raving fans”. They are those people who act irrationally. They exhibit addictive, compulsive or obsessive behavior patterns. Like a teenager in love, they cannot entertain the thought of an alternative. There can be only one love.

In my capacity as Co-Founder and General Partner in Follow[the]Seed, a new breed of venture investment fund, I’ve developed the Raving Fans™ model, with the goal to identify killer habit forming products and services in the post-seed, pre-A round stage. Our research showed that in the vast majority of the last decade successes, “raving fans” were a driving force, and arguably the reason why Facebook has bought Instagram and Whatsapp, and has offered to acquire SnapChat. (A short intro to the Raving Fans™ can be found here http://followtheseed.vc/2015/08/whats-the-story-behind-the-ravingfans-model/)

Usually, most users are not raving fans. That’s why they are harder to spot and that’s why we need non-aggregated data. Since this data is not usually available from data providers such as AppAnnie, SimilarWeb, Flurry and Google Analytics, we’re currently using an SDK — a piece of code that can tell us that a unique individual has either logged in or out of a specific platform.

This enables us to handle queries such as “what percentage of daily active users are using the product before 8AM?” or “what percentage of the bi-weekly active users are using the product more than 3 times a day?”. These are part of a set of almost 200 queries that we can ask — and generally designed to answer 3 questions.

1. Does this product have a critical mass of Raving Fans?

2. Does this group of Raving Fans continues to grow over time?

3. Does this group have higher than category average retention rates?

The value of the model lies in knowing how to automatically answer these questions.

The bottom line for the Raving Fans™ model

Investors can ask entrepreneurs with MVPs/POCs with a few thousands of users to connect with the Raving Fans™ model. The data is then processed automatically, and within 14 day to 3 months they can have an invest recommendation. Meaning: The investors’s time will be well spend examining this opportunity with an “expensive” human talent.

Final word

Predictive analysis is just making it first steps in the VC industry. You can expect a healthy dose of skepticism. I don’t believe in one fits all solution — the raving fans is suitable for habit forming companies with a working product, not to all companies. It does not aim to replace the human manager, but rather gives it better tools for filtering and analysis.

It is one approach and I expect that the coming years will produce many others.

But at the very least, it is a tool to create more equality for entrepreneurs, and increase the likelihood of investors not missing out on great deals.