What Is a Session in GA4? Counts, Page Views and Users

April 13, 2026

Author: Shusaku Yosa
セッション数とは?PV数・ユーザー数との違いとGA4での確認方法をわかりやすく解説

A session groups a sequence of activity on a website or app. In GA4, a new session starts when an eligible interaction occurs without an active session. The default inactivity timeout is 30 minutes, but check the property’s configuration.

Opening another page does not necessarily start another session, and closing a browser tab does not always end one immediately. Page views describe displays, sessions describe grouped activity, and users describe identified visitors.

Compare the metrics

Metric

What it counts

Important distinction

Page views

Web page displays

Repeat displays can count again

Sessions

Groups of activity

One session can include several pages

Users

Identified users within a period

One user can have several sessions

The GA4 Views metric can include web page_view and app screen_view events depending on scope. Users are not exact real-world people. Do not assume a universal users ≤ sessions ≤ page views relationship across reports with different definitions and scopes.

A timeline example

Assume consistent identification in one browser, a 30-minute inactivity timeout and one correctly recorded page view per display. This is a fictional example.

Action

Cumulative page views

Cumulative sessions

Users

10:00: open article A

1

1

1

10:05: open article B

2

1

1

10:10: reload B

3

1

1

10:50: open C after 40 minutes without activity

4

2

1

The result is four page views, two sessions and one user. Switching devices or changing identification conditions can produce a different user count for the same person.

Understand session boundaries

In GA4, crossing midnight does not by itself force an ongoing session to restart. A campaign change during an active session also does not automatically start a new one. Avoid importing Universal Analytics rules without checking the GA4 definition.

Returning in the same browser before timeout can remain part of the same session even after closing a tab. Conversely, leaving a page open does not prevent inactivity timeout.

Why session_start is not always the session total

session_start records the start of a session, but the Sessions metric should not always be equated with that event’s count. GA4 uses session identifiers and, in relevant reports, an efficient estimation method.

A session ID alone can occur for different users. Detailed reconstruction needs the appropriate user identifier as well. When comparing a warehouse calculation with a standard report, align period boundaries, identity and processing assumptions. An estimate is not guaranteed always to be lower than a separately calculated total.

Read sessions in GA4

Use Traffic acquisition for session source, medium or channel. Use Landing page to analyze where visits begin. Align dates, web/app scope, hostname and filters before comparing reports.

Pages and screens includes pages visited later in a journey, so its views are not interchangeable with entry sessions. If a required metric is missing, an authorized editor can customize the appropriate detailed report.

Diagnose changes before acting

Pattern

First checks

Sudden increase or decline

Tags, consent, filters and releases

Journey breaks at booking

Cross-domain configuration and identifiers

Ad clicks do not match

Different counting units, repeat clicks and collection gaps

Visits rise without useful outcomes

Intent, landing experience and outcome quality

More sessions are not automatically better. A broken booking flow can force repeat visits without creating value. Combine acquisition volume with engagement and meaningful outcomes, and record measurement changes alongside marketing changes.

Keep marketing reporting consistent

Use consistent metric definitions and reporting periods when reviewing results. See NeX-Ray’s supported integrations and reporting features, or start for free.

References

Official references checked September 28, 2026. Interface labels and available features can vary by account and rollout.

Related posts