Over the course of the last decade, digitization of the world economy has intensified and accelerated, disrupting all corners of industry. Incumbents with longstanding market dominance were overturned by seemingly unthreatening companies: start-ups.
These start-ups embraced technology and agility to reshape entire value chains. The disruptive speed of change brought by start-ups has been nothing short of a revolution; sometimes described as theFourth Industrial revolution. For us, the underlying revolution is a revolution of scale.
Scale can be defined as growth which is enabled by machines, softwares or products that function without requiring human input. If the last two centuries have been about maximizing the ability for humans to perform more productively, this one is about removing the limiting human factor and enabling machines to become truly autonomous.
Some industries were ripe for this revolution; transportation, consumer finance or entertainment to cite obvious examples. Others have, for the moment, resisted as a result of higher barriers to entry, relying more on networks, relationships and acquired knowledge. Management consulting is such an example; one whereby the value of trust and interpersonal relationships, combined with complex knowledge frameworks, have thus far proven beyond machines. Nevertheless, even these barriers are now being scaled by companies like Palantir. Their model of combining consulting services and scalable products is reshaping the services of the major consultancy firms and the evolution of the whole industry.
We believe that investment in early-stage companies, otherwise known as venture investment, is on the precipice of undergoing a revolution of scale. Let’s begin by looking at venture investment from the top down. Cambridge Associates data on early-stage investments shows that venture capital as an asset class has out performed public markets over the past decade. The figures also show that venture capital performance is poorly correlated with public market performance.
Data from AngelList on early-stage investor styles shows that success was directly correlated with the size of their portfolio. Simplistically speaking, the more companies invested in, the greater the chance of getting it right. Overall venture fund performance often boiled down to one successful company whose outsized returns compensated for the majority which failed.
Taken together, these two datasets suggest investing in early stage companies is more of a “spray and pray” approach than investing in later stage companies, particularly publicly listed companies; also that, if we leave aside the exceptional venture investors, it is a pure probability game for those who can afford the risk.
This casino-type approach has become more prevalent and was well-described recently by Sam Altman, the former president of the world’s leading seed money start-up accelerator, Ycombinator. SamAltman, who is now the CEO of OpenAI, tweeted in Dec 2020:
I have never seen early-stage startup investing be less disciplined than it is right now.
Many of the current players are looking at the past to explain why and how start-ups are succeeding. However, none have yet been able to see those companies coming. Equally, while some companies claim their ability to predict future scenarios, none are doing so by identifying companies we have not heard of.
We think that the reasons for this are the result of how data is being gathered, structured and understood. Existing solutions to find start-ups rely heavily on non-scalable data gathering. This often results in static information, updated manually and infrequently, that is either quickly out-of-date or prone to manipulation and misinterpretation. The data is also often structured according to predefined and unfit classification frameworks, which means that it is backward looking and is slow to show how technology is changing the dynamics of an industry. Tesla, for example, would be classified by most data providers as an automotive company, which, to us, misses the core point that Tesla is as much a software company as anything else.
We want to go beyond databases and matchmaking platforms, to build an infrastructure to uncover the hidden forces that shape early stage investments. We believe this infrastructure will make it possible to approach investments in start-ups in a much more structured manner.
The waters in which start-ups are operating are becoming increasingly visible through data. Social media data, recruitment data, market data are increasingly available and provide vital context with which to understand holistically how a start-up is faring, who its competition is and how rapidly the relevant market is growing. As the world becomes increasingly digital, something the pandemic has only accelerated, the greater the data picture is becoming and with it, the clearer the view of what is happening for a company.
Data is of course only the raw material; the real value is in the ability to understand it. For us at nr2, as we worked with leading institutions, corporations and investors to make sense of the data for hundreds of thousands of start-ups, we quickly faced the need to develop algorithms to contextualize and compare them. It also raised challenging questions around how to identify new concepts or industries. This work led to the development of a novel algorithm, which was not pretrained, that has enabled us to cluster companies that would otherwise never have been correctly associated. It means that even as new sub-sectors, or indeed new industries emerge, we can continue finding the most promising companies and understand the broader trends.
We deeply believe that we cannot, nor should even try to, predict the future. Technological disruption is inherently unpredictable and black swan-like. However, through intelligent data collection and analysis, we have seen that it is possible to see, at a very early stage, tectonic shifts that were emerging. This has led us to believe we can create a scalable knowledge tool to understand trends.
Data and its intelligent use is key during the revolution of scale. To harness its value, we are building a scalable search engine. The search engine leverages both structured and unstructured sources, including news articles, recruitment websites, social media and more. This enables a dynamic understanding of the start-up ecosystem that reflects what is happening, in close to real-time, without pre-conceptions that might cloud judgement.
The addition of unstructured resources not only improves the overall resolution of the picture, but more importantly provides the necessary tools to mitigate cultural bias. And culture matters. If you want to know which companies are doing well in different geographies, then it requires gathering data in the local language and understanding the tools through which information is shared. This meant that building Chinese and Korean language algorithms took over half a year respectively to develop, but the result is we can take a local approach to understanding the Chinese and South Korean markets; the first two geographies we have successfully mapped.
A search engine is not only a tool to find information, it is also a sensor. By understanding what investors, entrepreneurs, students or recruiters are looking for, we can see traction before it becomes obvious and use it to identify unpredictable trends.
To date, with only 5,000 users, we have been able to identify trends in artificial meat, transportation and identify fundraises before they became public. As usage grows, we will be able to identify and share with the community of search engine users an increasing breadth of trends specific to early stage ventures.This will revolutionize how we understand early stage investing.
It will not mean we can predict which company will prevail, but by identifying trends and emerging technologies, we will be able to provide all of our users a structured way to identify the most promising start-ups around the world.
China and South Korea are both significant economies with extremely promising start-ups, high digital penetration rates and significant language barriers, making them ideal for our initial focus. Additionally, we have extensive experience living and working in the region which has made it easier to understand the evolution of information, how it has changed and how it is effecting global change.
A significant time investment was required to find a way to gather and index structured and unstructured information in the local languages from both ecosystems. This investment is now delivering. Today, we are able to process directly in Chinese and Korean in a way that means we can see local trends and index information based on relevant terms. This includes being able to account for local concepts which do not translate directly; terms like 黑科技 (literal meaning ‘black technology’; non-literally meaning ‘deep tech’) which are important in Chinese descriptions and easily lost in translation. This publicly available data is then classified, enabling us to index start-ups for search.
We are breaking knowledge and language barriers that have historically curtailed understanding of start-ups. What we can make available for search by keywords today, we will make available through natural language tomorrow.
We will deliver the means to find and understand, in any local market worldwide, which fast-growing new companies are emerging and what they are doing with both the micro and macro market dynamics accounted for.
All seamlessly through a single intelligent search engine.