Using Machine Learning to Delight Users

I talk through various ways we use machine learning to surprise and delight customers at Groupon, and how we put in place frameworks to do so at scale in production systems. I lay out 6 key principles for developing machine-learning based products.

Transcript

1. USING MACHINE LEARNING TO DELIGHT USERS Laura Hamilton Group Product Manager at Groupon Product Management Festival November 15, 2018
2. 6 principles of developing machine-learning based products
3. CHALLENGE: GROUPON SCALE 48.8 million active customers globally Nearly 1.5 billion Groupons sold More than $30 billion saved by consumers 190 million app downloads Top 6 ranked app of all time Tens of billions of user actions per month Decisions made in fractions of a second - 15 countries and 500+ markets 1 million merchants worked with to date
4. TIMELINE 2008 Groupon founded It all began with the “daily deal.” SmartDeals First version of deal-ranking algorithm deployed. 2010 Feynman Personalized rankings based on user click, purchase, and impression history. 2015 Click model Search results dynamically learned based on click-through-rate 2016 2017 Yoda Scalable ML framework for search and recommendations 2018 QED ETL management platform with built-in cleaning and anomaly detection Flux Machine learning assembly line allows testing multiple models and deploying rapidly Chatbots Quickly and scalably respond to customers’ needs with Natural Language Processing Supply intelligence Predict the performance of each merchant that could run on Groupon Logistics Move inventory in advance of forecasted demand
5. ML APPLICATIONS Discovery and personalization—Provide personalized deal recommendations based on your previous purchase and browse history. Supply Intelligence—There are millions of merchants we could call at any time to get onto our platform. How do we pick the best ones? Fraud prevention—Fighting the bad guys in real time. Logistics—Reduce the time it takes customers to receive their Groupon Goods by sending extra inventory to our warehouse in advance of forecasted demand. Image recognition—Typing is so 2017. Take a picture of your credit card and we’ll do the rest.
6. Principle #1: Replace “business rules” with algorithms
7. Principle #1: Replace “business rules” with algorithms
8. Principle #2: Use data to supercharge decision-making.
9. Principle #2: Use data to supercharge decision-making.
10. Principle #3: A personalized experience is table stakes
11. Principle #3: A personalized experience is table stakes
12. Principle #4: Treat data as a first-class citizen Credit: XKCD
13. Principle #4: Treat data as a first-class citizen.
14. Principle #5: Maintain a clear separation of concerns.
15. Principle #6: Invest in data. Data is your moat.
16. Principle #6: Invest in data. Data is your moat.
17. Questions?