GA4 Dimensions vs. Metrics: Scope, Examples, and Reporting
Published:
Last Updated:
Category: Web Analytics
Authors: Shusaku Yosa
A GA4 dimension defines how data is classified; a metric measures a quantity or rate within that classification. The distinction is about analytical role, not whether a value looks like text or a number. This guide uses official documentation checked on October 11, 2026 to help you choose reporting pairs and avoid misleading totals.
Distinguish a dimension from a metric by its role
The official introduction to dimensions and metrics provides the starting distinction. Page title is a dimension and views is a metric. An hour or numeric category label can still be a dimension because its role is to classify observations.
Example | Role | Interpretation |
|---|---|---|
Page title | Dimension | Check whether several pages share a title |
Hour, such as 09 or 10 | Dimension | A time category, not a quantity to add |
Session source | Dimension | A classification of session acquisition |
Sessions | Metric | Not interchangeable with users or events |
Event count | Metric | Occurrences, not verified people |
Engagement rate | Metric | A rate with a defined numerator and denominator |
Avoid memorizing a supposedly permanent number of dimensions. Look up the current definition of the relevant field. Similar labels can refer to different units or collection moments.
Turn a business question into a reporting pair
Write the question first, then choose the grouping, metric, date range, host, page scope, and acquisition conditions. A report should answer a defined question rather than accumulate every available column.
Question | Candidate grouping | Measure and conditions |
|---|---|---|
Which sources start visits? | Session source / medium | Sessions; specify organic-only or all channels |
Which pages begin visits? | Landing page | Sessions; fix production host and intended paths |
Which events arrive? | Event name | Event count; verify the event’s meaning separately |
Where were users first acquired? | First user source / medium | Select an appropriate acquisition metric, such as new users |
All-channel sessions cannot substitute for organic landing sessions. Views across every page also differ from visits beginning on an article. Save the cohort and site-wide results with their distinct scopes.
Choose compatible user, session, and event scopes
GA4 scopes include user, session, and event. First-user acquisition, session acquisition, and event-related attribution answer different questions even when their labels contain the word source.
The objective is not to prohibit every combination of scopes. Check compatibility and analytical meaning. A combination being available in an interface or API does not prove that it answers your business question. Align field definitions, dates, filters, and identity conditions when comparing reports.
Reporting identity also affects how activity is associated with users. Do not automatically equate an analytics user measure with verified people, especially when cross-device recognition is incomplete.
Build a consistent table in reports or Explorations
- Select the intended property and fix dates, production host, and page scope.
- Choose the relevant primary dimension and metrics in the report.
- Record any secondary dimension, comparison, or filter.
- In an Exploration, select the corresponding fields and place them in rows, values, and filters.
- Investigate differing totals by checking definitions, scopes, filters, and display limitations.
More detail is not always more useful. Adding many dimensions to sparse data creates small rows that are hard to interpret. Start with the most relevant grouping and add another only to answer a specific follow-up question.
Design a custom definition only when needed
Prefer a predefined field when it answers the question. Otherwise, separate collection of an attribute from registration of a reporting definition. Follow Google’s custom-definition guidance and choose the appropriate scope.
Decision | Example or check |
|---|---|
Question and values | Define a maintained set of editorial categories |
Scope | Decide whether the attribute belongs to an event, user, or item |
Collection and definition | Verify the parameter or property is sent, then register the matching definition |
Timing | Check reporting availability, normally 24–48 hours after collection and registration |
Value management | Avoid unnecessarily unique IDs or URLs as categories |
Privacy | Do not send names or email addresses as attribute values |
Registration is not a promise that all historical data will become available under the new attribute. Record collection and registration dates and value changes. Follow PII-prevention guidance rather than creating custom categories simply to expose user or session identifiers.
Read user totals, weighted rates, and missing values correctly
In a fictional example, category A contains illustrative users 1, 2, and 3; B contains 3 and 4. The row counts are three and two, but the combined unique count is four, not five. These labels explain overlap; they are not a recommendation to send personal identifiers to GA4.
Rates also require their denominators. Suppose A has 40 engaged sessions out of 100, and B has 10 out of 20. Their rates are 40% and 50%. The combined rate is 50 divided by 120, approximately 41.7%, rather than the unweighted average of 45%.
Demographic buckets such as 18–24 and 25–34 cannot establish an exact count of people in their twenties. Consider Google signals, collection coverage, unknown values, and privacy limits before generalizing to everyone. Investigate missing values through collection and field applicability. When reporting business outcomes, separate event occurrences from verified successful people.
Continue with the relevant guide
