Kevin Systrom and Mike Krieger – How to Build a Great Product

Episode Summary

Executive Summary: Patrick O'Shaughnessy interviews Instagram co-founders Kevin Systrom and Mike Krieger about product strategy, machine learning, growth, and leadership. They reflect on building Instagram from a tiny startup into a billion-user platform, and on what they learned scaling inside Facebook.

Main Topics: Machine learning as practical leverage (Priority: 5/5): They see ML as useful but not magical: powerful when data quality and problem fit are real. Product philosophy and jobs to be done (Priority: 5/5): Instagram succeeded by solving clear user jobs: better photos, faster sharing, and speed. Scaling from startup to global platform (Priority: 4/5): They describe the shift from physical infrastructure and manual ops to cloud, open source, and scale. Community-first decision making (Priority: 5/5): They argue that products must serve multiple communities while staying authentic and principled. Growth, distribution, and product evolution (Priority: 5/5): They discuss how stories, ranking, and friend-focused usage changed Instagram's engagement and behavior. Management lessons from Facebook (Priority: 4/5): They learned to separate management from individual contribution and to keep product work central. Investing, curiosity, and future waves (Priority: 3/5): They are exploring investing and new technologies while staying pragmatic and learner-minded.

Key Arguments: ML works best where data is rich and the problem is well-posed, not as magic. They built Instagram by solving three jobs: photo quality, sharing, and speed. Open source plus AWS made startup infrastructure vastly cheaper and easier to scale. Founders should look at raw data and edge cases; hidden product use reveals real needs. Community-first means serving users authentically, not optimizing for founder ego. Stories succeeded because Instagram's original feed got too selective for everyday sharing. Product changes should be bold and reversible, not half-launched and timid. Management should be a separate career track, but product building must remain central.

Data Points: Instagram monthly users: over a billion people use it every month - Kevin describes the scale by the time they left Instagram team size at departure: over a thousand people - Kevin describes the size of the organization when they left Number of offices: three main ones and probably six or seven other ones - Kevin describes Instagram's geographic footprint Founding team size at launch: two people - They say they launched Instagram as just Kevin and Mike Early team size after first hires: four people - They added Josh and Shane after launch Launch spend: $65K total by launch day - Kevin recalls Instagram's launch cost Funding raised: $500,000 raised - Kevin compares launch spend to seed funding Growth pivot year for Stories: 2016 - Mike says Stories was launched broadly in that year Early company size when Facebook deal closed: 16 - Mike says they were at 16 people during the transition Team growth after Facebook: almost 80 - Mike describes growth about a year later Management count: probably 12 managers - Mike describes how the org structured management Machine-learning data point: a million examples - Mike mentions model training scale when discussing feature engineering Internship company size: maybe eight people - Kevin describes Odeo when he interned User behavior on Instagram Direct change: a billion people - Kevin notes that even a small percentage became many users at scale

Pivotal Quotes: "machine learning. It's just math." — Instagram engineer cited by Kevin: Used to emphasize humility and realism about ML "We may not be right, but we're not confused." — Case study quote referenced by Kevin and Mike: Captures their conviction without overconfidence "do the simple thing first" — Instagram value: Describes their product and engineering philosophy

Implications: Their next chapter is still undefined, but the playbook is clear: stay close to real user pain, treat new tech as a tool, and keep learning fast.