Data Fortification at MailChimp:
Trends, Pitfalls, and the Customer Experience
June 2015
Largest email service provider
8.5m users, 15b sends, 5b actions
350 employees all in Atlanta
Email is almost 40x better at acquiring new
customers than Facebook and Twitter -
McKinsey & Company (2014)
55% of compani...
Improve the product
Improve the user experience
Classic Research
Forecast and Optimize Processes
Inform Customers
Contextualize (Respond)
Trigger
Predict and Personalize
Predict and Personalize
Predict and Personalize
Danger 1: Everything is a nail
Danger 2: Tinkering
Danger 3: When your audience isn’t your customer
Danger 4: Locking your scientists behind a fence
Danger 5: Building a better carriage
Danger 6: The Content Problem
Danger 7: Over-complexity
Danger 8: Hidden consequences
1
1. Know what’s possible
• Data
• Techniques
• Technologies
2. Identify opportunities for
the business
• In constant conversation
• Not all problems are
data problems
3. Choose what solves the
problem
• Avoid added complexity
• Use internal data where
possible
• Think about the UX
Early Middle Late
Come back for the usability, stay for the data
“We want to mine our own data!”
BARF
Pro Reporting (start with classic research)
• In-app data-mining across any set of
emails
• Composite analysis
• Comparati...
XXXXXXXXXXXXX
XXXXXXXX
Choose the tools to fit the job
• Apache Kafka – Moving high volumes of
data into the reporting system in real
time, fault...
Thanks!
@John4man
NASSCOM Big Data and Analytic Summit 2015: Master class: Big Data Trends and Pitfalls in Email Marketing and Beyond
NASSCOM Big Data and Analytic Summit 2015: Master class: Big Data Trends and Pitfalls in Email Marketing and Beyond
NASSCOM Big Data and Analytic Summit 2015: Master class: Big Data Trends and Pitfalls in Email Marketing and Beyond
NASSCOM Big Data and Analytic Summit 2015: Master class: Big Data Trends and Pitfalls in Email Marketing and Beyond
NASSCOM Big Data and Analytic Summit 2015: Master class: Big Data Trends and Pitfalls in Email Marketing and Beyond
NASSCOM Big Data and Analytic Summit 2015: Master class: Big Data Trends and Pitfalls in Email Marketing and Beyond
NASSCOM Big Data and Analytic Summit 2015: Master class: Big Data Trends and Pitfalls in Email Marketing and Beyond
NASSCOM Big Data and Analytic Summit 2015: Master class: Big Data Trends and Pitfalls in Email Marketing and Beyond
NASSCOM Big Data and Analytic Summit 2015: Master class: Big Data Trends and Pitfalls in Email Marketing and Beyond
NASSCOM Big Data and Analytic Summit 2015: Master class: Big Data Trends and Pitfalls in Email Marketing and Beyond
NASSCOM Big Data and Analytic Summit 2015: Master class: Big Data Trends and Pitfalls in Email Marketing and Beyond
NASSCOM Big Data and Analytic Summit 2015: Master class: Big Data Trends and Pitfalls in Email Marketing and Beyond
NASSCOM Big Data and Analytic Summit 2015: Master class: Big Data Trends and Pitfalls in Email Marketing and Beyond
NASSCOM Big Data and Analytic Summit 2015: Master class: Big Data Trends and Pitfalls in Email Marketing and Beyond
of 44

NASSCOM Big Data and Analytic Summit 2015: Master class: Big Data Trends and Pitfalls in Email Marketing and Beyond

John Foreman, Chief Data Scientist, Mail Chimp
Published on: Mar 5, 2016
Published in: Presentations & Public Speaking      
Source: www.slideshare.net


Transcripts - NASSCOM Big Data and Analytic Summit 2015: Master class: Big Data Trends and Pitfalls in Email Marketing and Beyond

  • 1. Data Fortification at MailChimp: Trends, Pitfalls, and the Customer Experience June 2015
  • 2. Largest email service provider 8.5m users, 15b sends, 5b actions 350 employees all in Atlanta
  • 3. Email is almost 40x better at acquiring new customers than Facebook and Twitter - McKinsey & Company (2014) 55% of companies generate more than 10 percent of sales from email - Econsultancy "Emailmarketing census” (2014)
  • 4. Improve the product Improve the user experience
  • 5. Classic Research
  • 6. Forecast and Optimize Processes
  • 7. Inform Customers
  • 8. Contextualize (Respond)
  • 9. Trigger
  • 10. Predict and Personalize
  • 11. Predict and Personalize
  • 12. Predict and Personalize
  • 13. Danger 1: Everything is a nail
  • 14. Danger 2: Tinkering
  • 15. Danger 3: When your audience isn’t your customer
  • 16. Danger 4: Locking your scientists behind a fence
  • 17. Danger 5: Building a better carriage
  • 18. Danger 6: The Content Problem
  • 19. Danger 7: Over-complexity
  • 20. Danger 8: Hidden consequences 1
  • 21. 1. Know what’s possible • Data • Techniques • Technologies
  • 22. 2. Identify opportunities for the business • In constant conversation • Not all problems are data problems
  • 23. 3. Choose what solves the problem • Avoid added complexity • Use internal data where possible • Think about the UX
  • 24. Early Middle Late Come back for the usability, stay for the data
  • 25. “We want to mine our own data!” BARF
  • 26. Pro Reporting (start with classic research) • In-app data-mining across any set of emails • Composite analysis • Comparative analysis • Post-hoc segmentation and exploration • Ability to share What’s working? What’s not? How do I do better?
  • 27. XXXXXXXXXXXXX
  • 28. XXXXXXXX
  • 29. Choose the tools to fit the job • Apache Kafka – Moving high volumes of data into the reporting system in real time, fault tolerant • Postgres – Our data is already structured, indexes on what we need (geo, etc.) • MLE – Easy to debug during v1 and understand why data is driving output • Reporting is seamlessly integrated into MailChimp proper reusing same JS libraries and PHP framework
  • 30. Thanks! @John4man

Related Documents