GA4 completely changed how we think about website and app users. Instead of just counting every visitor like Universal Analytics (UA) did, it focuses on real engagement. To truly master GA4, you have to go beyond the basics and understand how it defines, measures, and reports on users, especially when you dive into the raw data in BigQuery.
The Core GA4 User Metrics
GA4’s data model is built around events, and users are defined as the people who trigger them. The biggest shift from UA is the emphasis on Active Users as the primary metric.
| Metric | Definition | Notes |
|---|---|---|
| Active Users | Unique users with an engaged session or new users within the timeframe | Filters out short, non-interactive visits; primary metric in GA4 reports |
| Total Users | All unique users who triggered at least one event | Includes both active and inactive users; always higher than Active Users |
| New Users | Users with their first-ever session (first_visit / first_open) | Can overlap with Returning Users if analyzing multiple days/weeks |
| Returning Users | Users who have had a previous session | Counts only users who had prior activity in your date range |
⚠️ Note: A single user can be both New and Returning in the same date range (e.g., first visit Monday, returns Wednesday).
GA4 Reporting Identity: Connecting the Dots
GA4 tracks a single user across devices and platforms using a multi-layered identity system, which is more advanced than UA’s reliance on a single browser cookie.
| Identity Space | How It Works | Best Use |
|---|---|---|
| User ID | Unique ID assigned to logged-in users | Track cross-device behavior; best for de-duplicated user analysis |
| Google Signals | Uses Google account login info and ad personalization consent | Helps GA4 deduplicate users across devices without login |
| Device ID | Client ID (web) or app instance ID (mobile) | Default fallback; counts same person on different devices separately |
| Modeling | Machine learning fills gaps for users who decline cookies | Estimates behavior based on similar consenting users |
In GA4 Admin, you can prioritize identity spaces: Blended (default), Observed, or Device-based. Best practice: use Blended and implement User ID whenever possible.
Diving Deep: GA4 Users in BigQuery
For advanced analysis, the GA4 interface isn’t enough—you need raw, event-level data in bigquery.
Note: if you haven’t started tour bigquery journey, start here: How to set up BigQuery linkage in your Google Analytics 4 property
user_pseudo_id (Default Identifier)
An anonymous ID tied to a user’s device and browser. Use COUNT(DISTINCT user_pseudo_id) for general user counts.
user_id (Cross-Device Key)
Populated if you implement User ID. Key to cross-device analysis and connecting GA4 data with internal systems.
By combining these IDs in SQL queries, you can analyze logged-in vs. anonymous users, cross-device behavior, and even replicate GA4’s Active Users metric with custom filters.
Implementation and Best Practices
| Practice | Reason |
|---|---|
| Always Implement User ID | Ensures true, de-duplicated cross-device user counts |
| Understand Metrics | GA4 counts may be lower than UA but reflect real engagement |
| Use GA4 Explorations | Segment, path, and funnel analysis for advanced insights |
| Combine GA4 with BigQuery | Enables 360° user view, cohort analysis, and CRM integration |
TL;DR
- GA4 focuses on engaged users, not all visitors.
- Active Users is the primary metric; Total Users, New Users, and Returning Users provide context.
- GA4 uses a multi-layered identity system: User ID, Google Signals, Device ID, and Modeling.
- In BigQuery, use user_id and user_pseudo_id for granular, de-duplicated analysis.
- Best practices: implement User ID, use Blended identity, match GA4 filters and date ranges, and combine GA4 with BigQuery for advanced analysis.
- For SQL implementation and raw data analysis, see our deep dive into GA4 users in BigQuery.
Leave a comment