How to Build an Engagement Scoring System in Mixpanel

Identify three to five behavioral dimensions that constitute engagement in your product. Common dimensions include: Frequency (how often the user returns), Depth (how many features they use per session), Breadth (how many different features they use over time), Collaboration (actions involving other users), and Recency (how recently they were active). Choose dimensions that map to the behaviors your most successful users exhibit, not just the behaviors that are easiest to track.

• Analyze your power users in Mixpanel — what behaviors distinguish the top 10% of users from the average?
• Validate each dimension by checking its correlation with retention: dimensions that don't predict retention aren't worth scoring

For each dimension, select the specific Mixpanel events that represent it. Frequency might be measured by 'session_start' count; Depth by the count of distinct event types per session; Collaboration by 'invite_sent' or 'comment_posted' events. Assign weights to each dimension based on its importance to your product's success — if collaboration is the strongest predictor of retention, weight it higher than frequency. The total weights should sum to 100.

• Use Mixpanel's Signal report to statistically identify which events are most correlated with your retention metric
• Start with equal weights and adjust based on data — over-engineering weights before you have validation data leads to false precision

Create a custom formula in Mixpanel's Insights report that computes the engagement score. For each dimension, normalize the metric to a 0-100 scale using percentile ranking against your user base. Then apply the weights: Score = (Frequency_percentile * 0.30) + (Depth_percentile * 0.25) + (Breadth_percentile * 0.20) + (Collaboration_percentile * 0.15) + (Recency_percentile * 0.10). This produces a single score between 0 and 100 for each user.

• Use Mixpanel's Formulas feature to combine multiple metrics into a single computed value
• Calculate the score on a rolling 14-day or 30-day window so it reflects recent behavior rather than lifetime activity

Segment users into engagement tiers based on their scores: Power Users (score 80-100), Engaged (60-79), Casual (30-59), and At-Risk (0-29). Create these as behavioral cohorts in Mixpanel that update dynamically. These tiers become the foundation for targeted analysis — you can compare feature adoption, conversion rates, and retention across tiers to understand what distinguishes each group.

• Validate tier boundaries by checking that the distribution roughly matches what you observe qualitatively about your user base
• Create a Mixpanel Dashboard showing the user count in each tier with a weekly trend line to track how the distribution shifts over time

The most actionable insight from engagement scoring isn't the current distribution — it's the movement between tiers. Track how many users migrate from one tier to another each week. Which users are moving from Engaged to At-Risk? What changed in their behavior? Which Casual users are becoming Power Users and what trigger events preceded the shift? Use Mixpanel's Flows report to visualize these migration patterns and identify the moments that matter most.

• Create a weekly comparison chart showing tier distribution this week vs. last week to spot sudden shifts
• Build a cohort analysis of users who moved from At-Risk to Engaged to identify what re-engagement actions worked

Validate your engagement score by correlating it with business outcomes: retention (do higher-scored users retain longer?), revenue (do they convert or expand at higher rates?), and referral (do they invite more users?). If your engagement score doesn't correlate with at least one business outcome, it's measuring activity, not engagement. Use Mixpanel's Retention and Funnel reports filtered by engagement tier to prove the connection.

• Build a retention curve per engagement tier — you should see clear separation between tiers
• Calculate the LTV difference between tiers to quantify the business value of moving users up one tier