June 2019 Summaries

Prachi Wadekar, a Product Manager at Mixpanel, discusses the phased launch approach for new features, emphasizing the importance of customer feedback in product development. Drawing inspiration from Amazon's customer obsession principle, Mixpanel prioritizes solving customer problems to create mutual value. The process involves launching a minimum viable product (MVP), collecting customer feedback, and iterating through a beta phase before making the product generally available. Betas are used not for quality assurance but to validate solutions and gather feedback, with private and public betas offering different levels of interaction. By building a pool of engaged users, Mixpanel leverages customer insights to refine products and ensure they meet user needs. The article stresses the significance of setting clear exit criteria for betas and maintaining a data-driven approach to track progress and success metrics. The ultimate goal is to achieve product-market fit, with a focus on continuous innovation and customer-centric problem-solving.

MAU, or Mobile Apps Unlocked, is an annual event in Las Vegas that gathers over 2,500 leaders from top mobile brands globally. At the event, Gil Sadis, VP of Product at Lemonade, shared how his team leveraged customer data to grow their user base by 500% in three years, showcasing Lemonade as a prime example of data-driven decision-making and innovation in the insurance industry. Lemonade, known for its AI-powered platform and unicorn status, uses tools like Mixpanel to analyze customer behavior, set priorities, and foster a culture of data-informed decision-making across all levels of the company. This approach has enabled them to innovate rapidly, eschew traditional product roadmaps, and focus on solutions that meet genuine market needs, contributing to their success and highlighting the importance of customer insights in driving strategic initiatives.

Mixpanel has introduced Query-Time Sampling as a solution to the challenges posed by their previous method of ingestion-time behavioral sampling, which was used to achieve fast and cost-effective analytics at scale. While ingestion-time sampling helped maintain rapid analytics by sampling data at the user level, it led to significant drawbacks such as data loss, trust erosion, reduced messaging reach, weakened A/B test significance, and irreversible data loss. In contrast, Query-Time Sampling retains all data and allows users to toggle sampling on or off during query execution, providing both speed and accuracy without data compromise. This new approach is supported by Mixpanel's core infrastructure enhancements, including a database migration to Google Cloud and a decoupling of storage from compute resources, aligning with their user-centric pricing model to support scalable analytics. Query-Time Sampling will soon be available for eligible Enterprise customers, offering a significant advancement in data analytics capabilities.