AI Within Venture Capital

January 2, 2022

What are the challenges?

For investment-based businesses, such as Venture Capital (VC), lead generation typically means discovering companies that meet certain criteria or tracking them until they do, then funnelling them through a structured assessment pipeline. What these criteria are depends on the type of market. For example, a VC firm focusing on Seed Stage will have very different criteria to one focussing on Series B. Whatever the criteria, the general process is usually very similar. It typically breaks down into 2 phases:

  • First, you need to discover candidate leads from the universe of potential companies. This typically involves searching & filtering across a range of data points, applying industry knowledge to identify candidate leads, creating watchlists, and reviewing these periodically until some match a specific set of criteria.
  • Second, validated leads are added to a pipeline, and investors make contact, form relationships, gather information, and progress the companies through the pipeline until they either invest in, drop, or park them in case they’re promising but not ready.

Traditionally investors have relied on their “gut feel” to make these kinds of investment decisions. A methodology prone to bias, inefficiency, and suboptimal outcomes. However, we’re seeing that the more technology-enabled VC firms are increasingly turning to data, advanced analytics, and AI to assist with and automate these decisions, especially in the earlier stages of the investment process. [1]

A whole ecosystem of digital tools have been built to address both of the phases, usually focusing on one or the other. The first phase is heavily dependent on having easy access to good quality data from a wide range of sources, and smart tools to help navigate the data. To accommodate this need many data aggregators have appeared to address different segments of the market, for example S&P Capital IQ, BvD FAME, and Bloomberg for publicly traded and more established private companies, or Pitchbook, Crunchbase, and Dealroom focusing on start-ups and scale-ups. The second phase requires a well structured CRM-like workflow. Tools for this include Salesforce for general-purpose pipelines, and Affinity which is more specialised and is popular with VC firms.

Although these tools provide access to a wealth of data and useful user interfaces, most of them don’t provide advanced analytics or AI to extract the most out of this data. For those looking to get ahead this is not a challenge but an opportunity. After all, getting access to good quality data is usually the most painful part of developing an analytics or AI capability. The “AI bit” can actually start adding value relatively quickly if you know what you’re doing.

Both of the phases mentioned above have their challenges, but I’d like to focus on the first phase, i.e. the discovery and tracking of leads from the universe of companies, and specifically how to be the best at it. Here, being better than your competitors means having access to more data, better analytics & AI, and more efficient tools. It all comes down to whether you can be the first to find the most promising companies so that you can be the first to place an offer. Anything that saves you time, allows you to focus, or provides more visibility is going to help you achieve this goal.

Using a publicly available SaaS platform, such as Crunchbase, is by definition accessible to anyone, including your competitors. You’re therefore competing on how much time you spend and how regularly you check these tools. You can of course gain access to more data by subscribing to multiple platforms, but this eventually becomes unmanageable.

How can we overcome these issues?

The logical solution to gain the edge without scaling human resources is to leverage the existing platforms by collating their data into a central data store, and investing into your own, customised tools, analytics, and AI that enables you to efficiently explore and analyse all of this data in one place. Then, when you decide that a new data source is needed, for example a news source, social media, talent movement, or internal data, you don’t have to adopt yet another platform. You simply need to integrate this incremental additional source into the mix.

This also enables you to perform analytics based on a fusion of data points from a wide variety of sources, which is important because a good lead is defined by a variety of factors. You can also take this one step further by generating customised lead scores (a metric indicating the likelihood of you being interested in a company) by encoding industry knowledge and experience into machine learning models using both labelled data and rules-based heuristics. All of this can save a huge amount of time by allowing you to ignore irrelevant leads and focus on applying your industry experience on just the bits that data and machine learning just can’t make a confident judgement on (yet).

How can we help you?

Unfortunately, developing all of these capabilities would require experience in product development, software engineering, data engineering, machine learning, as well as knowledge of the industry and data landscape. It would also require a considerable up front financial investment. Having worked with VC firms and professional services, we at Curvestone saw this as an opportunity. Given that we already had years of experience in developing data-centric and machine learning products we decided to invest in developing the foundations and building blocks for a “pre-structured-pipeline” company discovery and tracking platform (yes, we need a better name). Our approach has always been consulting-led which means that our outputs are always customised and unique to the needs of our clients to give them the edge they need to compete. If you’re interested to learn more about these challenges and opportunities then feel free to reach out.


July 6, 2023
Read more
June 23, 2023
Read more
June 7, 2023
Read more