Page 70 - Forbes - India (January 2020)
P. 70
Interview
(he declines to divulge details), there are positive surprises and are geographic concentrations.
Rajaraman, in an interview with sometimes negative ones. People Most venture capitalists spend their
Forbes India, talks about the fund’s can say they saw a great market and time cultivating their network.
data-driven approach to investment, went in, but that’s all hindsight. But entrepreneurship and
the need for an investor to identify an Our median valuation for Series innovation have gone completely
addressable market and lessons learnt A has been $20-25 million. At that global. How do you invest in these
from his India investments. Excerpts: valuation, we can take those risks. At hot companies across the globe? The
the later stages, however, investors are answer that we hit upon was, the
Q How has online retail played out evaluating market size and price of the right way to do this is to use data
in India vis-à-vis the US? How many round etc. People coming in later have and algorithms. If we can build a
such businesses, both horizontal and to be more cautious on the pricing. big data set of startups in the world,
vertical, can India accommodate? There is always a little bit of a herd then we can run machine learning
The US and India are not different. mentality in all venture capital. It models on it and we might be able to
In the US, there is one leader happened in the US as well. A lot of identify these companies and instead
in ecommerce, Amazon. In it is the fear of missing out. Suppose of waiting for them to contact us,
India, there are two businesses I don’t have a self-driving car startup we can proactively reach out and
competing, Amazon and Flipkart. and they become big, I will look invest. We license data from many
India as a market can possibly foolish. All these are eventually baked data providers, like Tracxn in India.
support one or probably two into the economics of the model, at We also crawl the web, we look at
big ecommerce companies. The least for the good funds. The bad funds social media and it’s all automated.
market is just not big enough. go out of business. When the boom At Rocketship, we probably have
There was a bit of over-investment starts, a lot of investors are drawn in the most comprehensive data set of
and the valuations were out of sync startup activity in the world—more
with reality. You can expect that than 10 million companies globally.
anywhere in venture capital. In certain “In India, things
70 verticals, there is room to build. There always take longer Q Is dependence on machine
was a belief that certain verticals will than you think. The learning a full proof method to
go online before India was actually identify potential winners? Are
ready for them, furniture for example. regulations change, the algorithms evolved enough to
India still needs an online-offline so companies have identify, for instance, fudged data?
model for some of these things and to adapt to those The first set of (machine learning)
that obviously takes longer and more models is a screen. The initial models,
capital to build. From a broad market things, which of the millions of companies in the
point of view, if anything can be doesn’t happen as data set, identify about 500 companies
done purely online, you can expect often in the US.” a year. We talk to them and get the
the market leaders like Amazon or second level of data from them and
Flipkart to be dominating those. then we invest in 10 of those.
There is some element to this that
Q Did we overestimate the market is data-driven and some element
in India? As an investor, how do that is our expertise. The algorithms
you differentiate between overall from the fringes who don’t necessarily will say whether the pattern looks
market and addressable market? have the expertise and when the bust interesting, but whether it is fudged or
I imagine so. We are an early-stage happens, a lot of these people leave. not is something that we will have to
investor. At that stage, it is hard to look and see. The important thing to
know the exact size of the market Q With Rocketship, was realise is that it is not the algorithm or
opportunity. We see if the business there any particular gap that machine learning model making the
has product market fit, whether the you set out to address? decision. It’s a combination of those
early cohorts look good and if there Both Venky (Harinarayan) and I models and us as venture capitalists
could potentially be a big market. understand venture capital and we with 20 years of experience. We
When I invested in Facebook, I had had our wins and losses. We knew are combining these things. I don’t
no idea what the size was. I believe the old model and didn’t really want believe in that (completely automated
even Mark Zuckerberg or Accel to do that again with Rocketship. decision-making) model. My strong
(another early investor) didn’t know The old model is a network-driven belief is that the future is about
the size of the market. Sometimes model. It works well when there humans and artificial intelligence
foRbes IndIA • january 31, 2020

