Where Do Startup Ideas Come From?


Jeffrey Bussgang

June 4, 2024

And How Will AI Impact Startup Ideation?

I never viewed myself to be particularly good at coming up with startup ideas. Maybe that’s why I became a VC. In the case of the two startups I worked for, someone else had the idea and I focused my time and energy on turning the idea into a business.

Maybe that’s why I am fascinated with whether there can be a systematic way to generate great startup ideas. In some ways, great startup ideas can seem like they come out of nowhere. After studying the startup ideation process for the last few decades, I’ve concluded that, in reality, they emerge from three primary approaches:

  1. Personal Experience: Many successful startups are born from the founders’ own experiences and frustrations. When you encounter a problem firsthand, you gain a deep understanding of the pain points and can envision a solution that others might not see. For example, SQUIRE was founded after its founders experienced the hassles of booking and paying for barbershop services.
  2. Technology Breakthrough: Other successful startups are created out of a novel technology or innovation. When a new technology wave emerges, founders swarm around it to explore and discover new pain points and value propositions. For example, the founders of C16 Biosciences were inspired by advancements in synthetic biology to create more sustainable, lab-grown palm oil.
  3. Domain Expertise. Many successful startups are created by a domain expert who has obtained some “earned secret” about a market need or gap. By leveraging their inside knowledge and nuanced understanding of the market, they can craft a product that uniquely exploits this need. For example, the founders of Shippo were former e-commerce entrepreneurs who were frustrated with the hassles and complexity of shipping products to their customers, inspiring the creation of a cloud-based shipping software platform.

But what if you aren’t inspired by one of these three paths? What if you can’t come up with a brilliant idea?

For many of my entrepreneurial Harvard Business School students, that is their essential question. For the last few years, I’ve been teaching them one systematic method: Research-Driven Innovation (RDI).

Research-Driven Innovation

RDI is a technique pioneered by Scott Brady of Stanford University Graduate School of Business and Innovation Endeavors. It is a systematic process to identify gaps in the market to generate startup ideas.

By analyzing industry trends, customer behaviors, and technological advancements, you can spot new opportunities to create significant value. RDI is particularly useful for generating innovative ideas in complex or unfamiliar markets. It combines thorough market research, expert interviews, and data analysis to uncover non-obvious problems and develop viable solutions. The RDI process includes:

  1. Team Formation and Alignment: Assemble a diverse team with complementary skills and define your success criteria. Create an opportunity rubric that outlines the ideal attributes of the market or industry you want to explore.
  2. Bench-Level Analysis: Choose an initial focus area and become conversationally fluent in the space. Develop a market hypothesis and identify key interview targets. This phase is akin to conducting a comprehensive literature review in academic research.
  3. Research Synthesis and Evaluation: Conduct curiosity-driven interviews with industry insiders and gather data to test your hypotheses. Synthesize the insights to evaluate potential opportunities systematically. This phase helps refine your understanding of the market and identify the most promising problems to solve.

AI Turbocharge

Incorporating AI can enhance the startup ideation process by analyzing vast amounts of data, generating insights, and simulating customer interactions. For example, you can use AI to conduct market analysis, gather competitive intelligence, and create detailed customer personas.

Modern AI tools allow you to rapidly iterate on your ideas and validate them with real-world data. By using RDI, founders can move beyond intuition and anecdotal feedback, employing a rigorous, data-driven approach to uncover high-potential startup ideas.

There is a common misperception that genAI tools like ChatGPT are not good at creativity and therefore won’t be useful for startup ideation. A study done by Wharton and Cornell professors suggests that conventional wisdom is wrong on that point. In a paper titled, “Ideas Are a Dime a Dozen: LLMs for Idea Generation in Innovation,” they showed that ChatGPT4 was able to generate higher quality startup ideas as compared to their Wharton students. Importantly, they also showed that the vast majority of the best ideas in the pooled sample came from ChatGPT4.


We might be entering a golden age of startups — new AI tools make startup ideation more efficient and insightful, new AI tools allow for more rapid execution of startup experiments, and new AI tools allow for faster startup scaling. And, as I’ve written in the past, AI agents are soon going to be available to orchestrate all of these processes, turbocharging the founder’s vision and plan.

I smiled when I heard Sam Altman of OpenAI speculate that at some point in the future, thanks to the power of AI, there will be a startup unicorn created by a solo founder. I will wonder how they came up with their idea.

🤖 I write about building startups in the age of AI. Subscribe to the Experimentation Machine newsletter: https://buff.ly/3viAAGH