Pick One Idea and Go Deep

John says founders cannot identify the best idea in the abstract; they need customer feedback to learn what works. He also warns against treating a lack of deep industry experience as a reason not to start, citing Blake Scholl, who moved from ad tech to founding Boom Supersonic.

Working on several ideas at once can produce weak evidence, leading founders to abandon promising ideas or stick with poor ones. John recommends choosing one, stopping work on the others, and committing so fully that even the company’s name, emails, and mission may change. He gives GovDash as an example: after at least five pivots, the team focused on government procurement and found strong demand.

John’s test for deep customer understanding is whether founders could run the customer’s business or teach a class about the problem. For a voice service product for cleaning companies, that means knowing the owners’ daily crises, the cost of missed calls, and what they would pay to prevent those losses. He recommends cycling between customer learning and product delivery, rather than waiting to finish all research before writing code.

John names three qualities to watch for: the product sits at the edge of current model capabilities, it can own an outcome rather than just sell software, and it pursues an ambitious version of the opportunity. He points to Corgi Insurance, which aimed to own the commercial insurance stack and acquired an insurance carrier during its YC batch. He argues that ownership of outcomes, trust, and regulatory permissions can matter as software becomes cheaper to produce.

If an idea fails after a deep commitment, founders can leave with clear customer evidence and stronger grounds for a pivot. John says deeper customer work often reveals structural problems beneath the surface-level pain points founders first noticed, including unmet needs around model limitations. His closing advice is to choose a direction and move quickly because focused work produces more information than dabbling.