Stacked Data Podcast 038 - How Flo run 1500+ experiments a year to drive product development
Nov 12, 2025
Dmitry Zolotukhin, VP of Analytics at Flo, leads a 30-person analytics team, driving a data-centric culture in product development. He shares insights on running over 1,500 experiments a year without sacrificing quality. Dmitry discusses how Flo transitioned from small tests to a robust experimentation framework. He emphasizes the importance of team education, process standardization, and balancing speed with trusted results. Dmitry also touches on future innovations like AI-driven decisions and meta-analysis to further enhance experimentation.
AI Snips
Chapters
Transcript
Episode notes
Experimentation Accelerates And Clarifies Decisions
- Experimentation gives both velocity and accuracy to product decision-making.
- Flo runs ~200 experiments simultaneously to parallelize progress and improve estimates of impact.
Early Wins With Content And Payments
- Flo began A/B testing in 2019 on content and its health assistant dialogues to increase completions.
- Early wins on content and payment screens convinced teams to adopt experimentation broadly.
Train Teams And Get Leadership To Test
- Educate teams continuously about experimentation concepts and tailor examples by domain.
- Get leadership to replace opinion with 'let's test it' and set clear team objectives to push adoption.