What to do when some metrics are up and some down? How to design statistically sound A/B tests?
How can a data scientist figure out whether a new product or feature should be tested? How to predict its impact?
How to create actionable metrics? How to guess their distribution? How to deal with sudden spikes/drops?
What to do if FB likes is up, but comments and time spent down? How to test on many metrics and deal with conflicting results?
Why large companies run A/B tests slowly increasing test traffic percentage?
How to test different prices without pissing off users?
What was the data-driven hypothesis behind FB stories, Netflix free trial, or Youtube Premium?
Should you try to increase the number of ads shown on a page? What about changing their spots?
Should FB merge Whatsapp and Messenger?
Conversion rate is down, but absolute number of conversions is up. Is it a good thing? When yes and when no?
How to guess the distribution of common tech company metrics?
What analysis would you do after a bug was discovered and fixed? How does this change if it was a product bug vs a logging bug?
What’s the difference between these new product case studies and the old ones?
Data science has evolved over the past few years. One of the main changes is that companies have started tracking tons of metrics at the same time.
This leads to many situations where trade-offs need to be evaluated, especially when some key metrics are up and some down. And this happens extremely often when choosing if launching or testing a new feature, as well as when designing an A/B test.
These new product case studies focus a lot on how to deal with those situations.
Also, testing strategies have become more refined from a statistical standpoint. Some of the new questions focus more on the statistical side of A/B testing, i.e. Gaussian approximation, confidence interval formulas, sequential testing, etc.
Are these new questions only included in the full course?
Yes, the only place where these new questions can be found is the full course in product data science.
They cannot be purchased separately from the course.