Upload
mongodb
View
1.494
Download
0
Embed Size (px)
DESCRIPTION
Citation preview
Actionable Analytics
Mongo Philly 2011Sheraton Society Hill
Robert J. MooreCEO, RJMetricsApril 26, 2011
What We’ll Explore
• My Background (Who is this guy?)
• Metrics & Developers
• Storing the Right Data
• Six Key Metrics
What We Won’t
• A Commercial for RJMetrics
• An In-Depth Technical Review
• A One-Way Lecture
Who is this Guy?
Robert J. Moore• Finance and Computer Science• Venture Capital Industry– Transition from Deal Sourcing to Data Analysis– Exposure to Tech Orgs of Amazing Companies
• RJMetrics– Technical co-founder and CEO– Hosted business intelligence– Providing access to deep insights for online SMBs
Metrics & Developers:Perfect Together
Developers Have Power
• Historically: power over product, progress, timelines…
• In the age of data: access to information
• Modern leaders “manage by metrics,” making those with access gatekeepers to success
A Growing Divide• As data sets get larger, they get farther out of
reach of non-technical data consumers in the enterprise
• Excel isn’t enough
• Access isn’t enough
• SQL isn’t enough!
A Gift and A Curse
• Developers become a key part of the business
• New technology can raise barriers before it lowers them
• Things get lost in translation
Embrace the Power• Know “what” and “why”
• Invest time in understanding the motivation behind data-related requests
• You will save time and add value in the long run
The Data
Good Practices• A database can be both functional and well-
suited for analysis (or warehousing)
• Overwrites are usually a bad idea
• Enforce consistency/cleanliness
• Timestamps are our friends
Common Themes
• Every business has its own unique needs
• Most operational data has common themes:– Entities (users, customers, visitors)– Actions of Value (purchases, logins, interactions)
The Metrics
1. Long-Term Engagement• Focusing on “total registered users” or “total
customers” is a common trap
• What happens to these users over time?
• What is your “Active” base?
• This is a common input to valuations
1. Long-Term Engagement
2. Repeat vs. First-Time Actions
• Digging deeper, we differentiate between newcomers and repeaters
• Acquisition vs. retention
• Helps separate biases from #1 caused by explosive new user growth
2. Repeat vs. First-Time Actions
3. Time Between Actions
• Actual magnitude can vary wildly by industry
• Ultimately, it’s the relative numbers that are interesting
• Does your product/service have “addictive” properties
3. Time Between Actions
Bias Warning
• Always consider the timeframe of the data you’re examining, especially when looking at metrics involving time
• Why might “average time between purchases” for newer customers look different than for older ones?
4. Repeat Action Probability• The “subsequent action funnel”
• Historically speaking, once someone has done something once, what is the chance they’ll do it again?
• Calling this a “probability” assumes it incorporates enough history to be representative of the long-term behavior of the population
4. Repeat Action Probability
5. Customer Lifetime Value• A key “actionable” metric– Informs marketing spend– Influences retention strategy
• Multiple Definitions– Lifetime Revenue (“Value So Far”)– Expected Lifetime Revenue– Lifetime Gross Margin (“Contribution”)
5. Customer Lifetime Value
• Segmentation Opportunities– Which segment are performing well?– Demographics– Geographics– Acquisition Sources– Behavioral Characteristics– Time-based Cohorts
6. Cohort Analysis
• The venture investor’s favorite slide• Incorporates everything we’ve discussed– Engagement– New & Repeat Actions– Timing of Events– Repeat Frequency/Probability– Lifetime Value Accumulation
6. Cohort Analysis• Pulling the data– Associate every event with two timestamps:
• The timestamp of the event• The “cohort timestamp” of the user responsible (this
can be a registration date, first action date, etc) – the value of this field will not change from record to record for the same user
– Break the users into “cohorts”• Yearly• Quarterly• Monthly• Weekly• Daily
6. Cohort Analysis
• Pulling the data (ctd)– Study these “cohorts” side-by-side, with their
“ages” on the x-axis instead of actual calendar dates
– This allows you to study how different customer cohorts have interacted with your site over time
– Are newer cohorts stronger or weaker than older ones?
6. Cohort Analysis: Traditional
6. Cohort Analysis: Relative
6. Cohort Analysis: Relative
6. Cohort Analysis: Cumulative
6. Cohort Analysis: Avg/Member
6. Cohort Analysis: Avg/Member
Conclusions
Conclusions
• As the data grows, so does its importance and so does the power of its keepers
• Design with future analysis in mind
• Always understand the “why” behind requests and you’ll save time in the long run
PlugsTwitter:@RJMetrics@robertjmoore
Visit our Website:http://www.rjmetrics.com/
E-Mail Me:[email protected]
We are hiring!http://www.rjmetrics.com/jobs