A covid-19 winter is coming to startup investment hard and fast. VCs and startups need to reflect how business will reboot, come the spring. We will take a look at data-driven VC, a trend that was already accelerating pre-crisis, and we believe will transform the sector from now on.
The market for startup capital is broken
Venture capital returns have been very disparate. The pooled return rate (IRR) of the VC asset class has been around 15% 1 . However, most of these returns come from a few headline-grabbing funds2 . Meanwhile, middle-of-the-pack funds don’t manage to beat the public markets and compensate for the extra risk. Bottom-quartile funds don’t even return the money invested. Charts still show an improvement for later vintages, but most of it is based on record valuations that are yet to be realized. This is likely to be revised down after the covid-19 public markets crash. 3
Early stage investors – VCs in particular – have been a force for the data-driven disruption of many industries. However, their own investment process remains very much driven by human interactions: who you know, how well to present, expert opinions, and relies heavily on manual process (events, meetings, emails, excel, pdfs). There are good reasons for this, and human insight will continue to be central given the nature of the investments. But the covid-19 crisis is showing many F2F interactions can be replaced with tech-driven alternatives, without a major impact, and the added benefits or reaching beyond usual circles, and less “group think”4 that has led to overbidding for the same “hot” opportunities.
At the same time, for startups, frequency and quality of reporting has been a predictor of success 5 . However, they are discouraged by a similarly manual and painful reporting process.
More generally, the market for startup capital is in need of a revamp: On one side, startups sit on innovative ideas and starve for capital. In particular those outside the limelight or that don’t fit the stereotypes. On the other side, most investors don’t have access to growth opportunities, as companies go public later in the cycle 6 . This pool money is many times larger than the ‘elite’ money that gets invested in VC or angel deals 7 , and it remains unavailable to startups.
Alternative markets, crowdfunding and more recently ICOs, have tried to address this mismatch. However, they have failed to scale. A key reason is that risk has remained high, as most investors don’t have the data, the analytic capabilities, or just the available time to analyse the myriad of offerings. As a result, regulators have become rightly concerned, and these options have remained the preserve of the same smaller pools of “elite” money.
This is even more pressing now. The covid-19 crunch has meant investors that still have liquidity need to find the best opportunities among a much larger number of applications from cash-strapped companies than they have the bandwidth to evaluate. And they need to screen and qualify them without the tried-and-tested F2F methods.
In the past, new technologies like the telegraph and the telephone since the late 1800s, or information technology since the 1970s drove transparency and multiplied the analytic capacity of investors. This drove big expansions of investment flows 8. Data and AI will have a similar impact now on the startup capital markets.
Data and AI will transform startup capital markets
Other segments in Finance have taken the lead using data and AI. Quantitative and “Quantamental” hedge funds have been using data and AI-driven strategies for years. Lending has been using them for decades.9
In the case of early-stage investment, there have been blockers: The patchiness and private nature of data, the very dynamic innovation markets, and the smaller scale of players to afford the significant investments. Now, these constraints are fading: Analytic capabilities are cheaper, ubiquitous and ever more powerful, with AI and the cloud. Enormous amounts of data are being generated by the “digital exhaust”. We may soon have “through-the-cycle” data, to separate hype and fads that drive growth during booms from company characteristics associated to long-term success.
Pioneering VCs on both sides of the Atlantic have started building their own data and AI capabilities. For example, EQT-Ventures in Sweden developed a platform called Motherbrain. It consolidates several puiblic data sources, and algorithmically sources deals for its 500 mill EUR venture operation. Several of their best performing investments were sourced using it. Others, like InReach Capital in the UK, Signalfire and Social Capital in the US (and its descendents Capital-as-a-Services, Tribe Capital, etc.) get private data from applicant companies to make funding and portfolio decisions.10
Covid-19 has accelerated the trend. For example, the NFX’s FAST Seed programme offers seed funding of 1-2 mill. USD with a 9-days decision timeframe, with a data-driven, software-enabled process. Once tested in these circumstances, it will become more and more prevalent once we return to normality.
However, spending millions 11 in building and maintaining their own data and AI infrastructures and highly-paid data teams, is beyond the reach of most VC firms who live on a 2% management fee. And most of this spend goes into low-value data preparation activities (scraping, cleansing, matching), replicated across the industry and that add little in terms of proprietary insight.
In addition, the quality and quantity of data needed to train AI models are yet to become available. Investors can get high-quantity of public/static data from the likes of Crunchbase, Pitchbook, Tracxn, etc., which cover hundreds of thousands of companies. Or they can scrape the web themselves. These sources are very valuable, but are also expensive to maintain and keep the data current. Alternatively, investors can get high quality performance and cap-table private data directly from the companies. But this is limited to the few tens or hundreds of companies to which they have access (portfolio companies, companies applying for funds). And this through a manual and painful process, or building expensive 1-2-1 integrations.
Enter data-driven early-stage investment
First, reduce the pain and get the data flowing …
The first step towards a data-driven VC strategy is streamlining the collection of company data (funding applications, portfolio monitoring). Investor relations platforms like Investory take the pain away from these processes, creating a seamless managed flow of data between companies and their investors. Startups enter their cap tables, KPIs, monthly updates through an automated process, and investors get visibility across their portfolio.
Here it is critical to provide startups the motivation to consistently report their data . Firstly, simplifying the process provides the basic incentive: Through platforms like Investory, the effort that goes into preparing and distributing the reports is vastly reduced. But in addition, it can provide startups with visibility for their fundraising efforts: Reporting through the platform allows them to share it with one-click with a large number of potential investors, and manage the flow of information.
For investors, this also simplifies their own reporting processes: For example, Investory consolidates in one place all the information necessary to produce Portfolio Reports. We are building more automation in the generation of these reports, to generate additional cost efficiencies for funds.
… then, harness the power of data
The next step is data-driven predictive analytics. Platforms like Investory will be uniquely positioned to offer insights based on high-quality company-provided data.
All activities in the investing value chain can be enhanced with data AI: For Fundraising, data and AI can be used to identify LPs that are likely to invest in the fund. For portfolio construction, data and AI can help identify promising leads (outbound dealflow) in a systematic way, qualify, validate and benchmark incoming pitches (inbound dealflow), and streamline the due diligence process. Also, they can provide valuable insights for portfolio management, including alerts and signals about the portfolio companies, and most importantly, enable data-driven follow-on investment decisions . Finally, they can help identify high-likelihood acquirers to exit the investments. Relying more on data has become a need during the covid-19 pandemic due to the lack of F2F opportunities, and will continue afterwards once investors have had the opportunity to prove it works.
Leveraging SaaS platforms, Investors will be able to gain an edge from their proprietary data, without the need to spend millions in their own AI capabilities, and benefiting from insights derived from a much larger set of high-quality data than they could gather themselves. To guarantee data privacy and confidentiality, models are “trained” based on anonymised data. Then they are “run” to provide specific insights on each fund’s proprietary data. Think credit ratings, or image scanning for cancer detection: Models are “trained” based on very sensitive personal data. Then models are “run” on specific cases to predict creditworthiness, or the presence of cancer. Sensitive data remains confidential to each owner, who is able to benefit from his proprietary information while tapping insights learned from the pool.
Finally, human insight and interactions will continue to be central to the early-stage investment business. According to EQT ventures, data and AI augment the investment team “to increase the capacity of how many companies we can look at”. The technology improves the ability to scope out deals in the pipeline, focusing the human effort on the prospects with the highest chances. 12
At Investory, we are working towards building this vision. We are launching a Startup Fundraising Program to help match startups with investors startups during the covid-19 crisis. We will aim our next funding round at enhancing our value to investors and startups for the first step of the data-driven strategy (portfolio monitoring & reporting), and at setting the foundations for the future data-driven AI investment era.
Notes, References
1. Pooled VC asset class IRR was 14.2% in 2005-15, according to McKinsey Private Markets Review 2019, Exhibit B [back to article]
2. The source of the VC Performance Data in the chart is Pitchbook Benchmarks as of Q1-2019. See also Money Talks – Things you learn after 77 investment rounds, Gil Ben-Artzy,, slide 14: “Do VCs make investors expectations?”
[back to article]
3. The Economist, Apr 4th, 2020, Technology Startups are Headed for a Fall
SIFTED, March 16th, 2020, Fundraising during coronavirus: Are startups screwed?
[back to article]
4. The Economist, Apr 17th, 2019, The wave of unicorn IPOs reveals Silicon Valley’s groupthink [back to article]
5 . TechCrunch, Jan 7th, 2020, Study associates frequency and quality of reporting with startup success, [back to article]
6. “Exchanges, investors, regulators, business groups and politicians have all expressed worries that the ability to take stakes in young and potentially exciting companies is being disproportionately enjoyed by small groups of insiders and elite investors rather than the wider investing public.”, Financial Times. [back to article]
7. VC assets under managements are around 1 trill. USD globally, according to McKinsey Private Marekts Review 2020, Exhibit 10. This is a tiny fraction of total world financial assets, projected at 900 trill. USD by 2020 by Bain (Public Vs. Private Assets: The Big Switch, Feb 25th, 2019, Figuire 3.1) [back to article]
8. . Ranald C. Mitchie, The Global Securities Market, A history. Oxford University Press, 2006 [back to article]
9. Financial Times, October 17th, 2019, Why hedge fund managers are happy to let the machines take over.
Forbes, Dec 12th, 2019, Quantamental Investing: A Fuzzy Term That Describes An Inevitable Future
The Economist, Oct 5th, 2019, March of the machines – The stockmarket is now run by computers, algorithms and passive managers
Financial Times, Aug 16th 2019, How AI will change the way you manage your money [back to article]
10. . EQT’s own description of Motherbrain After their initial success, they have just launched a 2nd fund with EUR 660 mill., one of the largest in Europe.
TechCrunch, June 26, 2019, VCs double down on data-driven investment models, covers Social Capital, and its descendents Capital-as-a-Service, and Tribe Capital
Financial Times. Dec 12th, 2017, Artificial intelligence is guiding venture capital to start-ups, covers InReach Ventures, SignalFire and EQT Ventures.
McKinsey Quarterly, June 2017, A machine-learning approach to venture capital, covers Hone Capital.
[back to article]
11. . “It took two years and £5m in investment for InReach Ventures to create the software” . For SignalFire “It took eight years and tens of millions of dollars to build”. Signalfire says he spends over $10m a year, while Mr InReach plans to spend at least £1m annually. Financial Times. Dec 12th, 2017
[back to article].
12. . ZDNet, Feb 12th, 2019, How an AI ‘Motherbrain’ helps venture capitalists pick investments
[back to article].
Investory.io helps startups telling their stories to investors, enhancing their narratives with trusted data.
Would you like to tell your startup story to investors? Send your case to info@vtlabs.online.
Investories is the blog by investory.io all-in-one investor relations portfolio management platform. Give it a try for Free!
