6. How to Be a Better Data
Driven Product Manager
By Emile Saad, Sr. Product Manager
7. About Me
● Sr PM at Groupon, KnowledgeView (startup in Beirut, Lebanon)
● Product management experience in mobile apps and mobile marketing
● Background in Computer Science, MBA
8. Agenda
● Automate and Simplify to Own Your Metrics
● Plan Ahead for Experimentation and Analysis
● Trust Your Instincts
● Key Takeaways
9. You run across the CEO in the hallway. She asks you:
“How is your product doing?”
10. Possible Answers?
“It’s doing great!”
Another option: “Last week, we added a new feature. It’s currently driving X
additional users to the product, an increase of Y% from last month”
12. Why Automate Your Metrics?
● Know the ins and outs of your product metrics
● Identify new opportunities and issues faster and address them
● Free up time to focus on important tasks
● Be comfortable answering questions on the spot
13. Invest Time Upfront to Automate your Metrics
● Identify main metrics sources and dashboards
○ If none, create them!
■ Dashboards: Tableau, Chartio, Looker, Google Data Studio
■ Logging and monitoring tools (Google Analytics, Splunk, Elastic…)
● Reserve 10-15 minutes every morning to review your main metrics
● Focus on most important metrics for your product. Dive deep when necessary
Tip: When starting a new PM role or new product, use your initial learning
days to identify metrics and compile/build your dashboard
15. Start With a Plan
Start with a simple, 1 page plan to guide your analysis/experiment:
● Explain how this experimentation/analysis will be conducted
● Define what success will look like
● Clarify what to do in case of success/failure
16. Explain Your Approach
● A/B Test:
○ Treatment description
○ Number of users
○ Test duration
● Gradual Roll out:
○ Percent will start, percent increases, duration
● Pre/Post analysis:
○ What are the main metrics and how will they be interpreted?
● New product/MVP: interviews, landing page data, etc.
17. Define What Success Will Look Like
S.M.A.R.T. Goal: Specific, Measurable, Attainable, Realistic, Time Bound
This feature will improve buyer conversion rate
VS.
This new feature will increase buyer conversion rate by 5%, bringing in $100K
additional revenue annually
18. Clarify What Happens in Case of Success or Failure
● A simple decision matrix helps clear the picture in advance and plan for all
scenarios
● Failure is OK, as long as we know how to deal with it!
Decision Matrix Metric 1 Success Metric 1 No Change Metric 1 Failure
Metric 2 Success Roll out Roll out Rollback
Metric 2 No Change Roll out Roll out Rollback
Metric 2 Failure Re-Evaluate Rollback Rollback
19. Tips
● Keep your plan simple! (unless your company has specific policies)
● Beware of extremes:
○ Analysis paralysis
○ Oversimplification
● How you say it matters
○ Make sure you are using proper convention
21. Things Are Not Always What They Appear To Be
● The data will not always make sense
● Mistakes happen, even with the best analysts
● Confirmation bias can set us on the wrong path
● Accept that data will not always be available
22. What To Do When Things Look Too Good To Be True
When you feel too good about
your data
● Sometimes it’s good to do own due diligence
● Better to identify issues early, than later
● Challenge yourself to be a better PM
Large variations. e.g. 20%
increase, when expected 5%
● Dive deep: look at secondary metrics
● Look for outliers
● Review setup for mistakes
● Look for macro trends
When estimates given to you
seem unrealistic
● Review models and question assumptions
● Consider initial MVP instead of full product to
quickly validate
● Wear the customer hat
23. Experience Will Get You There!
Remember that it’s a marathon, not
a sprint. Continuous learnings from
daily metrics, launches, successes
and failures will make you much
better at suspecting when
something doesn’t look right
Identifying,
launching and
analyzing
Familiarizing
with metrics
and customer
trends
Learning from
customer
trends and the
data
24. Key Takeaways
1. Own your metrics and data by automating and simplifying
2. Plan ahead before experimenting
3. Trust your instincts and don’t be afraid to question the data