1.2.1 Applications companies vs. tooling companies

👱 Personal story 👱

After NVIDIA, I wanted to join an early-stage startup. I was considering two AI startups that were similar on the surface. Both had just raised seed rounds. Both had about 10 employees, most of these were engineers, and were ready for hypergrowth.

Startup A already had three customers, had just hired its first salesperson, and was ready to hire more salespeople to aggressively sell. Startup B had two customers, who they called design partners, and had no plan yet of hiring salespeople. I liked the work at startup B more but thought that startup A had a better sales prospect than startup B, and for early-stage startups, sales are essential for survival.

When I told this dilemma to a friend, who had invested in companies similar to both A and B, he pointed out to me that I forgot to consider the key difference between these two startups: A was an applications company while B was a tooling company.

Applications companies provide applications to solve a specific business problem such as business analytics or fraud detection. Tooling companies create tools to help companies build their own applications. Examples of tools are TensorFlow, Sparks, Airflow, Kubernetes.

In industry lingo, an application is said to target a vertical, whereas a tool targets a horizontal.

Applications are likely made to be used by subject matter experts who don’t want to be cumbered with engineering aspects (e.g. bankers who want applications for fraud detection or customer representatives who want applications to classify customer tickets). Tools are likely made to be used by developers. Some are explicitly known as devtools.

In the beginning, it’s easier to sell an application because it’s easier to see the immediate impact of an application and there’s less overhead for adopting an application. For example, you can tell a company that you can detect fraudulent transactions with 10% higher accuracy and the company can just input their data into your application and get out results whether a transaction is fraudulent or not.

For a company to adopt a tool, there’s a lot of engineering overhead. They might have to swap out their existing tool, integrate the new tool with the rest of their infrastructure, retrain their staff or replace their staff. Many companies want to wait until a tool proves its usefulness and stability to a large number of companies first before adopting it.

However, for tooling companies, selling becomes a lot easier later on. Once your tool has reached a critical mass with a sufficient number of engineers who are proficient in it and prefer it, other companies might just become a user without you having to sell it. However, it’s really, really hard to get to that critical mass for new tools, and therefore, in general, tooling companies have higher risks than applications.

After talking to my friend, I realized that it’s normal for a company like A to have more customers than a company like B early on. But it doesn’t mean that A has a better sales prospect. In fact, having two large companies as design partners is a really good sign for B.

A year later, both companies acquired a similar number of customers and grew to have around 30 employees, but more than half of company A are in sales whereas 80% of company B are in engineering.

This new understanding helped me narrow down my choices. Because I preferred building tools for developers and wanted to work for an engineering-first instead of a sales-first organization, the decision became much easier.

⚠ Ambiguity ⚠
Whether a company is an application company or a tooling company might just be a go-to-market strategy. For example, you have a new tool that can address use cases that companies aren’t aware of yet, and you know it’d be hard to convince companies to make significant changes to their existing infrastructures for uncertain use cases. So you come up with a compelling use case that can’t be done without your tool, build an application around that use case, and sell that application instead. Once customers are aware of the usefulness of your tool, you switch to selling the tool directly.

🌳 Tip 🌳
If you're unsure whether the role involves working with an application or a tool, here are a few questions you may ask.

• Who are the main users of your product?
• What are the use cases you're targeting?
• How many people does your company have? How many are engineers? How many are in sales?

This book was created by Chip Huyen with the help of wonderful friends. For feedback, errata, and suggestions, the author can be reached here. Copyright ©2021 Chip Huyen.