Must Have Key Business Analytics & Insights – Product Analytics

This is the second post from series of blog posts related to ˜Must Have Key Business Analytics and Insights˜, and this post will look into covering in-depth about product analytics.

Product centric analytics

When it comes to analytics consumption, product team is the first and foremost important business unit within a company; and they should have complete insights into the platform usage.

At high level it should help to understand:

• what is working, and what’s not
• what is most liked and the least
• what part of platform is most revenue generating
• where and how users are getting dropped (signups or feature usage)
• how users are getting engaged with the platform
• what devices and browsers are most used

These analytics helps them to A/B test on sections of the platform where it’s not working and helps to build new features that’s most liked by majority of the users. At the end of the day, product team mainly focuses on user experience and engagement, and partially works with marketing team on optimizing new user acquisition channel.

Even though it is product(s) and business logic dependent from company to company, here am presenting most common data points:

User engagement and behavioral histogram

This is the most important, one and only metric that every product engineer should have on day-to-day (if not real-time) with histogram (comparing over a period of time).

• how many users are coming to the platform
• how often users are coming back to the platform
• unique users
• new vs. existing users
  • usage
  • spend
  • engagement
  • behavior
• how much time in an average user is active on the platform
• classify & segment users behavior on
  • user activity
    • very active
    • active
    • moderate &
    • inactive
  • usage of each product category, application and/or core features
    • identify what has been used actively, how often and average usage time
    • identify low usage patterns
    • identify usage path drop-outs (funnel drop)
    • co-relate common multi usage patterns (more than 90% of users who uses product A also uses product C; only 5% of users who uses product B also uses product D)
    • new vs. existing
  • spend
    • high spend
    • medium spend
    • low spend
    • no spend
  • state
    • segment user state change, percentage
    • classify closed users and predict churn rate
      • segment active vs. inactive drops and predict what caused it
    • projected revenue drop/increase due to state change
  • predict
    • growth
    • revenue
    • usage
• user platform
  • device
  • browser, version
  • operating system and version
• cohort analysis (spend, usage, behavioral segments across all product features)
  • by user profile characteristics (age, gender, demographic, salary range, organization, job title)
  • by purchase type (personal, premium, enterprise, etc.)
  • by common usage patterns
  • by new vs. existing
  • by user platform (device, browser, OS, etc.)

Error rate

Real-time product related errors (all 4xx error codes) and error rate statistics across all product applications.

User journey analysis

This is yet another important feature and product team should know how the user journey looks like within the platform(products, features, pages, sections within a page, etc.):

• top landing and least used
• most & least spend
• most & least average time spent
• most & least drop-outs
• most & least common journey path (feature X to feature Y and then to Z)
• segmenting journey path based on user cohorts (80% of iPhone 5 commonly uses X, Y and z features, 40% of Mac with safari uses product A, etc.)

make sure to cover all possible critical user journey paths for all the above metrics within the platform through all possible cohort analysis.

Histogram is the key

All the above data points (being it usage or behavior) should have the capability to compare over a period of time (compare today vs. yesterday, this week vs. last week, this month vs. last month or over a period of last few years or months, how new vs. existing users behaving for last few months, how a particular cohort looks like in last 10 days, etc.).

Evolve over a period of time

Over time, product analytics should evolve by adding new data points as the product team adds new features or launches new version, or when drastic user behavioral changes are detected or when  specific A/B tests are carried.