How investors assess early-stage AI startups

Photo credit: Pixabay.

The way an AI startup is built—vertical (VAS) or horizontal (HAS)—allows investors to quickly assess a startup, think about the investment needed, and determine the type of questions to ask to get a deeper understanding of a company.

A VAS is focused on solving problems for a specific type of customer, while a HAS is science-driven, usually run by academics.

VAS startups usually need to show the usual business metrics for user/customer growth and defensibility, while HAS startups generally revolve around pure technology and/or a killer team. Investors make a snap judgment based on a startup’s framework and work backwards to approach an investment. In truth, VAS and HAS are more like two axes, X and Y, and a startup falls somewhere between them; but for simplicity, these two conditions are discrete.

More specifically, if a startup is a VAS, investors may want to wait to see user/customer growth metrics (if they are bullish) and even revenue (if they are bearish) before deciding. The only solution at that point for founders is to continue to drive growth and demonstrate that they can acquire users at a cost-effective rate (if the startup is B2C) or have commitments from large enterprise clients (B2B). All these are proxies for potential revenue, if one assumes a revenue number per user/customer.

Breaking this down further, the key to growth is to build a great product that your users love. You can use Net Promoter Score to monitor user satisfaction, survey a swath of users, or take advantage of the plethora of methods in the market to find how users really think. There isn’t really a panacea on how to build a product that users will love, but there is one principle on the ability to iterate quickly based on user feedback.

My favorite story is how Pinterest, initially to overcome slow growth, talked to its users face to face to figure out what to build and overcome its growth limitations. Some founders instinctively know what users want, but for many others, they’ll need to get used to regularly listening to their users to find out what they could not live without.

The three risks with VAS are: (1) the business can be small and might not generate the required return for investors, (2) the team does not have the required expertise to grow market share, and (3) there is no defensible moat.

For a HAS, investors generally want to see a technical validation signal. A signal could mean a key team member, advisor, early-stage investor, or potential investor who is technical enough to understand the technology and its potential. The reason for this is that most investors are not technical, and having a signal that validates the superiority of technology overcomes an investor’s cognitive bias on investing in something they do not know.

Founders should think about having a technical validation signal and actively enlist one. You may also make sure that the investors lined up are extremely technical and able to commit sizable capital before returns.

There is a larger perceived risk from doing a HAS since it is “all or nothing.” The things investors want to see are whether a startup is truly innovative and/or whether the technology provides a cost or use advantage that is 10 times greater than the incumbent. However, there is a big perception that only famous professors can create a HAS.

Thank you Eamonn Carey, Jeffrey Paine, Kenta Adachi, Shumpei Fukui, Naoki Kamimaeda, and Alexandre Winter for feedback.