Table of Contents
Mailchimp ranks among tech’s greatest bootstrapped exits. With almost no outside capital, the founders sold it for $12 billion, an “infinite IRR” outcome by any measure.
More impressive is how they did it: staying true to product-led growth and serving small-business customers. Ben Chestnut ran product while Dan Kurzius owned customer success, and they held that line from day one.
For a look inside their mindset, listen to Ben’s episode on the Grit podcast. His candor, paired with the calm sense that “this is enough,” stands out in a world where even the Valley’s hyper successful often keep chasing more.
Free vs. Freemium
Lost to the Internet archive is a series of blog posts Ben Chestnut wrote himself. One that every SaaS leader should track down is his reflection on Mailchimp’s path to Free versus Freemium. In a world where “freemium” too often means a paywall labyrinth, Ben’s approach still feels different:
We spent years building a powerful, affordable, profitable, self-serve product. We invested heavily in our API, which now tops 70,000 users. We mastered deliverability, scalability, and abuse prevention. Then the cloud made all of it even cheaper, so we passed the savings on and made the product free.
Omnivore, Spam Abuse, and Prevention
That focus on abuse prevention is another lesson the messaging world should study. Long before “AI” became a household term, Mailchimp built Omnivore, trained on an NVIDIA Tesla supercomputer, to spot bad actors inside legitimate-looking traffic.
Lack of permission, in an otherwise perfectly legitimate business email, is very subtle and much harder to detect.
The same principle applies to SMS: Permission gaps hide inside messages that look fine on the surface. Omnivore’s early machine-learning playbook is still a masterclass for today’s text-messaging platforms trying to balance reach with trust.
Finally
A cohort is formed by the shared experience of the technologies of that time.
We all cut our teeth on the early internet, SaaS, product-led growth, AI, and GPU-powered machine learning on NVIDIA cards, concepts we were testing long before LLMs became mainstream. The cohort’s results follow a classic Pareto curve, yet those shared experiences and lessons remain a deep well of practical insight.