What I Wish I Knew About Data For Startups

This is a presentation I made at MTL Data about the things I wish I knew when starting out at PasswordBox.

Video & Slides: http://www.jeannicholashould.com/talks

Transcript

1. What I Wish I Knew About Data For Startups Jean-Nicholas Hould Lead Data Analyst, Intel Security
2. In Summary Stuff I wish I had known Building a data culture Tactics
3. ~4 year timeline May 2015 Amazon Redshift, Looker January 2013 Joined PasswordBox (Employee #11) June 2013 Launched product December 2014 Intel Acquires PasswordBox May 2016 Processing 1B records per month August 2016 105 employees, 4 in data team number of events processed per month
4. Our Data Team Some tools we use and love: Dir. Data Engineering Co-founder PasswordBox Data Scientist Data Scientist Data Scientist
5. Organization Product Teams Data … iOS Android Desktop Backend … Support Teams
6. Three Themes Empower People Deliver Value Garbage in, Garbage out
7. Empower People tools & people share & communicate
8. Tools & People Tools are just tools Be a guide, not a gatekeeper Self-Service Tactic: Metrics Office Hours
9. Share & Communicate Share analysis, even if inconclusive. Make sure insights translate into actions. Communicate often. Tactic: Share data at team lunch / internal newsletter.
10. Deliver Value build vs buy ship often
11. Build vs Buy Always consider buying vs building. Building: Better for control and flexibility Buying: Cheaper in time and resource Tactic: Don’t build another dashboard from scratch. (Looker)
12. Ship Often Investments in engineering VS analysis Ship by small increments. Asana’s first data warehouse: MySQL Tactic: Start with basic infrastructure. Iterate.
13. Garbage In, Garbage Out document your data test your data control & own
14. Document Your Data Document actions and attributes tracked Teams should have access to documentation Machine-readable documentation Tactic: Document in excel. Transition to JSON/YAML
15. Test Your Data Implement a process for data quality Automate testing Less data, but better quality Tactic: Build internal tools to validate data quality. (Our internal tools to test tracking & reporting)
16. Control & Ownership Access to raw data for complex questions. Easily start/stop sending tracking. Tactic: Proxy your tracking and add new destinations with ease. (This is how we added Redshift)
17. Key Takeaways Empower People Be a guide not a gatekeeper. Share your data analysis often. Deliver Value Don’t build another dashboard from scratch. Start small. Iterate. Garbage in, Garbage out Invest in data quality: document, test, monitor. Own and control your data.
18. Reading List • How to build stable, accessible data infrastructure at a startup - Asana Blog • Everything We Wish We'd Known About Building Data Products - DJ Patil • Buffer’s Data Architecture - Buffer Blog • Facebook's Aha Moment Is Simpler Than You Think - Mode Analytics and a seamless plug…. • What I Wish I Knew About Data For Startups - jeannicholashould.com • Don't Build Another Dashboard From Scratch - jeannicholashould.com
19. s thank you jean-nicholas hould @nicholashould jeannicholashould.com