Most early-stage teams track too much or too little. Here's how to set up product analytics that actually informs your decisions — from your first 10 users to your first 10,000.
Our recommended product analytics platform
Amplitude: behavioral analytics, cohort analysis, funnel tracking, and AI-powered insights. Used by 3,000+ companies including Twitter, Dropbox, and Walmart.
The two failure modes: tracking everything (overwhelm, no signal) and tracking only vanity metrics (pageviews, signups) that feel good but don't drive decisions.
Product analytics done right gives you one thing: clarity on what behaviors correlate with retention, revenue, and growth. Everything else is noise.
Before you write a single line of tracking code, define the one metric that best captures the value your product delivers to users. This is your North Star Metric (NSM).
For your startup
Ask: "What is the action that, when a user does it, we know they're getting real value from our product?" That action is your North Star. Set up your analytics to measure it above everything else.
Events are the atomic unit of product analytics. Every meaningful user action should fire a named event with relevant properties. Good taxonomy makes analysis easy; bad taxonomy makes it impossible.
// Good: Specific, consistent naming convention
analytics.track("Project Created", {
project_type: "marketing",
template_used: true,
user_plan: "pro",
time_since_signup_days: 3
});
// Bad: Vague, inconsistent
analytics.track("click");
analytics.track("project_made");
analytics.track("NewProject");Use past tense for events
"Project Created" not "Create Project"
Use noun + past verb
"File Uploaded", "Payment Completed", "User Invited"
Add context properties
Plan type, user cohort, feature flag — you'll want to slice by these later
Be consistent
Pick a naming convention (Title Case, snake_case) and never deviate
Amplitude's free tier covers up to 50,000 monthly active users — enough for most early-stage products. Here's the fastest setup path:
Create account and project
Sign up at amplitude.com, create a new project for your app, and grab your API key.
Install the SDK
npm install @amplitude/analytics-browser (web) or the relevant mobile SDK. Initialize on app load with your API key.
Identify users
Call amplitude.setUserId(userId) on login. Add user properties: plan, signup date, company size, role.
Track your core events
Start with 5–10 events: sign_up, login, activation_event, key_feature_used, upgrade_initiated, churn_event. Add more only when needed.
Build your first 3 charts
North Star metric over time, activation funnel, Day 7 retention by signup cohort. These three charts will tell you everything important.
Activation Funnel
Shows where new users drop off before experiencing your product's value. Fix your biggest drop-off point and everything downstream improves.
Retention Curve
The shape of your retention curve tells you whether your product has product-market fit. A curve that flattens above 20–30% at Day 30 is a PMF signal.
Feature Adoption by Plan
Shows which features drive upgrading and which are ignored. Double down on what drives revenue, kill what doesn't.
Tracking before defining
Never start instrumenting without a written event taxonomy. Retroactive cleanup is 10x harder than getting it right first.
Treating all users the same
Always segment by cohort (signup date), plan, acquisition channel. Averages hide the most important signals.
Optimizing for top-of-funnel only
Signups are cheap to improve. Retention is hard. Obsess over Day 7 and Day 30 retention — that's where value lives.
Analytics without action
Every chart should have an owner and a threshold. "If activation drops below 40%, we investigate by next Friday." No metric without a decision process.
Amplitude
RecommendedProduct-led growth, behavioral analytics, cohort analysis. Best-in-class for understanding user behavior. Free up to 50K MAU.
Mixpanel
Strong alternativeSimilar to Amplitude. Stronger event-based querying UI. Slightly steeper learning curve. Free up to 20M events/month.
PostHog
Developer-friendlyOpen-source, self-hostable. Combines analytics, session recording, feature flags, and A/B testing. Good for teams with data sovereignty needs.
Google Analytics 4
Marketing onlyFree, widely used, but designed for marketing (sessions, traffic sources) not product behavior. Don't use as your primary product analytics tool.
Start building data-driven products today
Amplitude's free plan covers up to 50,000 MAU with unlimited events, behavioral charts, and cohort analysis — everything you need to understand your users.
See Amplitude DetailsI look at product analytics setups as part of due diligence at Value Add VC. The companies that can answer "what's your Day 30 retention by cohort?" instantly are the ones operating at a different level. For more on building data-driven products, see the Value Add VC blog.