How to Set Up Product Analytics in Amplitude: Step-by-Step Guide

Step 1: Create Your Account and Project

Sign up for Amplitude at amplitude.com. Create an organization and your first project. Most teams create separate projects for each product or platform (web app, mobile app, API). If you have a single product with a web and mobile client, start with one project and use platform as an event property to distinguish data sources.

Configure your project settings: set the time zone to match your primary user base, enable IP-based geolocation if relevant, and configure your data retention period. The free plan retains data indefinitely, but paid plans offer configurable retention windows.

Create a development project separate from your production project. Route events from your staging and development environments to the development project. This keeps your production analytics clean and gives engineers a safe space to test their instrumentation before shipping.

Step 2: Design Your Event Taxonomy

Before writing a single line of tracking code, design your event taxonomy. This is the naming convention and structure for all events your product will track. A well-designed taxonomy is the foundation of useful analytics. A poorly designed one creates confusion and technical debt that is expensive to fix later.

Use a consistent naming format. The most common conventions are “Object Action” (e.g., “Project Created,” “Dashboard Viewed,” “Report Exported”) or “Verb Noun” (e.g., “Created Project,” “Viewed Dashboard”). Choose one and enforce it consistently. “Object Action” is more common and groups related events together when sorted alphabetically.

For each event, define its properties (additional context sent with the event). A “Project Created” event might include properties like project_type, template_used, team_size, and source (where the user initiated the action from). Properties make events analytically rich without requiring separate events for every variation.

Document your taxonomy in a tracking plan spreadsheet or document. For each event, record: the event name, description, when it fires, its properties (name, type, expected values), and which platform(s) implement it. This document is the contract between product and engineering.

Step 3: Implement the SDK and Core Events

Install the Amplitude SDK in your application. Amplitude provides SDKs for JavaScript (browser), React Native, iOS (Swift), Android (Kotlin), Node.js, Python, and other platforms. For web applications, the JavaScript SDK is a few lines of initialization code.

Start by implementing user identification. Call Amplitude's identify method when a user signs up or logs in, passing their user ID and any user properties (plan type, role, company size, sign-up date). User properties enable segmentation and cohort analysis later.

Implement your core events first. Focus on the 15-20 events that map to your critical user journey: sign-up, onboarding steps, first key action (the “aha moment”), core feature usage, and conversion or upgrade events. Resist the urge to track everything at once. You can always add more events later, but removing noisy events from analysis is harder than adding missing ones.

Test your implementation in the development project. Use Amplitude's User Lookup feature to find your test user and verify that events are arriving with the correct names, properties, and values. Fix any issues before deploying to production.

Step 4: Set Up User Properties and Groups

User properties are attributes about the user (not the event) that enable segmentation. Common user properties include: plan_type (free, pro, enterprise), role (admin, member, viewer), company_size, industry, sign_up_date, and onboarding_completed (boolean).

For B2B products, set up Amplitude's group analytics. Groups let you analyze behavior at the account level (company, team, or workspace) rather than just the individual user level. This is critical for B2B products where buying decisions are made at the account level. Configure your group type (usually “Company” or “Workspace”) and assign users to groups during identification.

Set user properties at sign-up and update them when they change. For example, when a user upgrades from free to pro, update the plan_type property. This ensures segmentation and cohort analysis always uses current user attributes.

Step 5: Build Your Core Dashboards

Create dashboards that give your team a pulse on product health. Start with three foundational dashboards: a Product Overview dashboard, a Feature Adoption dashboard, and an Onboarding Funnel dashboard.

The Product Overview dashboard should include: daily and weekly active users (DAU and WAU), the DAU/MAU ratio (a measure of engagement intensity), new user sign-ups over time, and your core conversion metric (whatever event indicates a user has received value from your product).

The Feature Adoption dashboard tracks how individual features are performing. For each key feature, show: unique users per week, frequency of use (events per user), and adoption rate (percentage of active users who use the feature). This dashboard answers the question: “Are users finding and using what we built?”

The Onboarding Funnel dashboard visualizes the steps new users take from sign-up to activation. Use Amplitude's funnel analysis to see conversion rates between steps, median time between steps, and where users drop off. Break down the funnel by user segments (plan type, source, device) to identify which segments have the lowest conversion rates.

Step 6: Create Behavioral Cohorts

Behavioral cohorts are groups of users defined by their actions (or inactions) within your product. They are one of Amplitude's most powerful features because they let you compare how different behavioral segments perform.

Create cohorts for your key user segments: “Activated Users” (completed the core action within 7 days of sign-up), “Power Users” (performed the core action 10+ times in the last 30 days), “At-Risk Users” (were active 30 days ago but have not returned in 14 days), and “New Users This Week” (signed up in the last 7 days).

Use these cohorts as segments in your dashboards and analyses. For example, compare the feature adoption dashboard for “Power Users” vs. “At-Risk Users” to identify which features correlate with engagement. This analysis often reveals the behaviors that predict retention, which informs where to focus product investment.

Amplitude also supports syncing cohorts to other tools (email platforms, ad platforms, CRMs). This means you can target specific behavioral segments with personalized messaging or outreach based on their in-product behavior.

Step 7: Establish a Review Cadence

Analytics only creates value when people look at it and act on it. Establish a weekly analytics review as part of your product team's rhythm. During this review, walk through the core dashboards, discuss any anomalies or trends, and connect the data to current product decisions.

For each feature launch, define success metrics in advance and schedule a post-launch review at the 7-day and 30-day marks. Compare actual adoption against the targets set in the PRD. This closes the loop between planning and outcomes and builds a culture of data-informed decision-making.

Set up Amplitude alerts for critical metrics. If your activation rate drops below a threshold, if DAU declines by more than 10% week-over-week, or if a key funnel step degrades, you want to know immediately, not at the next scheduled review. Alerts ensure the team is responsive to significant changes without requiring constant dashboard monitoring.