Product analytics portfolio · Mixpanel

My Product Metrics

How I chose, mapped and set action thresholds for the product metrics of seven AI products, from agentic exploit validation to parametric crop insurance. Each chapter answers the same five questions, with screenshots of the live dashboard.

Mayukh Ghosh · Director of Product Management

7apps and boards

38metrics with thresholds

5questions per app

Threshold status across all metrics · synthetic

Read this first

Disclaimer: synthetic demo data

Metric values are for demo purposes only

Every number, trend and funnel on this site was generated using synthetic data scripts. None of it reflects real customers, real usage or real product performance.

Thresholds are proposed starting points, to be validated with real baselines.
Flat lines after mid-October show when the synthetic event feed stopped, not product behaviour.
Known data gaps are called out where they appear.

What is real

The product workflows, from each app's GitHub repo and the Agent Val workflow video
The event design and metric definitions
The Mixpanel dashboards and their structure

The portfolio

Seven apps, seven north-star metrics

Each card shows the app's north star and the status of each of its threshold metrics. Select a card to jump to its chapter.

Method

One method, applied to every app

Based on the product analytics approach from the session "Key Analytical Tools & Communicating Persuasively with Data", built in Mixpanel.

Reading guide

Five questions answered for each app

Threshold status

On trackInside the green band

WatchIn the amber band: monitor

Act nowIn the red band: take the listed action

Every chapter uses real screenshots of the app's v2 Mixpanel dashboard. Select a screenshot to enlarge it.

Across the portfolio

Where each app stands against its thresholds

Count of threshold metrics by status for each app (synthetic data).