Data and experimentation is becoming core of product development in todays world. In this talk I will talk about what I learned from building various products, why data and experimentation matter when building products used by millions of users. The talk will also touch on how we at Customator plan to make it easy for any one to follow a data driven product development approach.
Venue : NUS School of Computing Seminar
2. About Me
• Founder, Customator
• Data Engineer @ Viki, a Rakuten Company
• Research Intern @ Paypal
• Teaching Assistant @ SoC
• Software Engineer @ Rocketful Inc.
3. Fun stuff I did
Products I built
• Tuition Agency for Indian education boards
• FML NUS
• CorruptionTrak
• Devops as a Service
4. Overview
• Early Products and Lessons
Learnt
• Data Driven Product
Development
• What?
• Why?
• How?
• Data @ Viki
• Why?
• How?
• Impact?
• Customator
• Activity
27. Experimentation and
Personalisation made easy
• Cross Platform
• Experimentation in algorithms, not just UI
• Easy to use for non-technical users
• Auto personalisation
• Data Driven experimentation
29. • Lets pick a popular product of your choice
• Facebook? Quora?
• Brainstorm through the goals of the product
• What should we measure?
• How can we improve the product?
Its difficult to get money from Freshmen, specially when they are your friends.
Create a buzz around the products you build.
Helping people more difficult than making money.
People love seeing other people (specially young people) drive change.
Don’t work on something that doesn’t excite you enough to work on it full time.
Choose your team mates / co-founders very carefully.
I think my slides don’t strictly follow the outline, but its a general overview
DDPD seems like a big term thrown around in the industry a lot. Lets try to figure it out together.
Your very own time table builder, built by students from NUS.
Lets up the stakes.
Lets say we make it a product and sell it to all university students around the world.
How do we make money? - And we make money by a subscription service.
So we found the problem, we came to a solution. Now how do we know if its working?
Visits
Visitors
Average time per session
Demographics, location, time of day
Every dimension of information gives you more insight
Drop offs at different points in the application
What are we optimising for?
Heatmaps
Funnels
Retention rates
Feature improvements / New feature development
Measure how the changes are performing
Experiment? Personalize?
https://www.zopim.com/ example
So what is AB testing?
Bottom left was the clear winner
Split url testing
Drove the conversions up by 20%
Learning which variations work best for what audience and then auto targeting.