Learning path · Level 5 · Tutorial 3

Read GA4 Acquisition, Engagement, and Key-Event Reports

Interpret text-based GA4 acquisition, engagement, and key-event reports to explain observed performance. Separate defensible findings from attribution uncertainty and possible tracking problems.

  • Level: Intermediate
  • Duration: 22 min
  • 9 steps
Start tutorial Back to the path
Read GA4 Acquisition, Engagement, and Key-Event Reports

What you will go through

  1. Establish the Report Context Identify the context, dimensions, and metrics needed to interpret a GA4 report extract. 2 min
  2. Choose User or Traffic Acquisition Select User acquisition or Traffic acquisition by matching first-user and session-scoped questions to the correct report and dimension. 3 min
  3. Read Channels, Sources, and Campaigns 2 min
  4. Interpret Engagement Without Mixing Units Compare user and session units, calculate engagement rates, and interpret engagement time without treating it as proof of business success. 3 min
  5. Separate Key-Event Counts from Session Rates Separate recorded `generate_lead` occurrences from the proportion of sessions that contain the selected key event. 3 min
  6. Make Like-for-Like Comparisons Compare matched GA4 periods without confusing higher activity volume with a higher session outcome rate. 2 min
  7. State What Attribution Cannot Prove 3 min
  8. Triage Report-Quality Warning Signs Identify report-quality warning signs, frame them as hypotheses, and select evidence that can verify or reject them. 3 min
  9. Write an Evidence-Based Performance Note Produce a short, defensible interpretation of supplied acquisition, engagement, and key-event results. 3 min

What you will learn

  • Choose user acquisition or traffic acquisition reporting to answer a specific audience-origin question.
  • Compare engagement and key-event results using compatible date ranges, dimensions, and metric definitions.
  • Distinguish key-event occurrences from the proportion of sessions containing a key event.
  • Identify plausible data-quality problems and state what additional evidence would help verify them.
  • Write a short performance interpretation that distinguishes observed results, attribution limits, and a justified follow-up check.

Before you start

  • Plan GA4 Events and Key Events
  • Create Consistent UTM Campaign Links
  • Set Marketing Objectives and Calculate Core KPIs

Step 1 of 9

Establish the Report Context

Identify the context, dimensions, and metrics needed to interpret a GA4 report extract.

Identify what the report is describing

Dimensions describe; metrics quantify

A dimension names what each report row describes. A metric is the quantity reported for that row.

For example, a row dimension might group activity by a channel label. Columns such as Sessions or Key events are metrics: they quantify activity associated with each row.

Read the dimension first, then read the exact metric labels. Numbers without those two pieces can be misleading.

Read the row label before the numbers

A schematic report table distinguishes one category column that describes rows from several numeric columns that measure activity.

The left-side dimension tells you what a row is about; the metric columns tell you how much activity is associated with it.

Capture the reporting context

Context makes a number interpretable

Before interpreting a GA4 extract, record its reporting context:

  • Property and relevant stream scope
  • Date range and property time zone
  • Any filters or comparisons
  • The row dimension and its scope
  • The exact metric labels and selected event, where relevant

For Harbor Workshops, use complete 14-day periods in UTC: 01–14 Apr 2026 and 18–31 Mar 2026. The case uses the same web-stream scope and no additional filters.

Tip

Comparisons are not experiments

A comparison selects subsets of observed data, such as one date range versus another. It does not create experimental control groups. A difference in report values can be observed; its cause still needs evidence.

Tip

Optional live-account practice

You can complete every activity in this tutorial from the supplied text extracts. If you choose to inspect a learner-owned GA4 property, standard reports normally require authorized access such as Viewer access or higher. Report collections and navigation can vary by property.

Spot what an extract does—and does not—tell you

Incomplete report extract

This practice extract deliberately omits its context note. It shows a row dimension and measured quantities, but not the property, dates, time zone, filters, or comparison setup.

Sessions with generate_lead is a supplied practice annotation. Do not assume it is a native column in every GA4 report.

Report rows

Selected channel rows from a report extract

The table provides values, but its missing context limits what can be concluded from them.
Default channel groupSessionsEngaged sessionsKey eventsSessions with generate_leadAverage engagement time per session
Organic Search1,000600403045 s
Email200140302060 s

Apply the report-reading checklist

Check the report context

Review the incomplete extract on the prior screen. Which response correctly identifies its structure and the context needed before making a reliable interpretation?

Step 2 of 9

Choose User or Traffic Acquisition

Select User acquisition or Traffic acquisition by matching first-user and session-scoped questions to the correct report and dimension.

Choose the scope before reading a row

Two acquisition questions

Acquisition reports can answer two different origin questions. Choose the report by the unit in the question:

  • User acquisition asks where recognized users were first acquired.
  • Traffic acquisition asks where sessions came from.

The source label may look the same in both reports, but its scope is different. Read the report name and the dimension before interpreting a row.

First acquisition versus session origin

A split diagram contrasts one person's first arrival from search with separate website sessions arriving from search and email.

Left: first-user scope. Right: session scope.

Report selection guide

Use the report and dimension that match the wording of the question.

Scope determines what a row means.
Question asks about…Use this reportScoped dimension
Where users were first acquiredUser acquisitionFirst user source / medium
Where a session originatedTraffic acquisitionSession source / medium

Apply the journey to each report

One user, two session origins

Follow one recognized Harbor Workshops visitor:

  1. Their first visit arrives through google / organic.
  2. Later, they return through the spring_workshops email campaign: newsletter / email.

In User acquisition, this person remains associated with their first acquisition source: google / organic.

In Traffic acquisition, the first session can be associated with google / organic, while the later return session is associated with newsletter / email. The Email channel describes that later session’s origin, not a replacement for the user's first acquisition source.

Recognized-user journey

One visitor first reaches a website through search and later returns through an email link.

A first-user view retains the original acquisition; a session view can show each arrival route.

Tip

Same label, different meaning

Do not swap these rows as if they answer the same question. A google / organic row under First user source / medium means “users first acquired from Google organic.” The same label under Session source / medium means “sessions that arrived from Google organic.”

Before comparing or describing an acquisition result, state whether it is first-user scope or session scope.

Practice: match the question to its scope

Select the matching scope

Match each Harbor Workshops question to the report or dimension that answers it.

Tap an item and then its matching pair.

Step 3 of 9

Read Channels, Sources, and Campaigns

Identify the acquisition label

Four distinct acquisition views

A source identifies the referring origin. A medium describes the traffic mechanism. A campaign identifies a marketing initiative when campaign information is available.

A default channel group is broader: GA4 applies rules to classify traffic into categories such as Organic Search and Email. It is not a campaign label that someone freely assigns.

Use the broad channel view to summarize traffic, then investigate with a compatible source / medium or campaign dimension when the question needs more detail.

From summary to detail

A broad traffic category branches into source, medium, and campaign details.

A channel group summarizes rule-based categories; source / medium and campaign views provide more specific acquisition detail.

Read labels without treating them as synonyms

Keep scope compatible

Keep the scope from the report you selected visible while changing detail. For a session-origin question, use the Session family of dimensions. For an original-acquisition question, use the corresponding First user family.

For the recognized Harbor Workshops visitor, google / organic can remain the first-user origin while a later session is newsletter / email. That later session can also carry the campaign spring_workshops. These labels describe different aspects of acquisition; they are not synonyms.

Parse the Harbor Workshops labels

Read the labels as separate pieces of information:

  • In google / organic, google is the source and organic is the medium.
  • In newsletter / email, newsletter is the source and email is the medium.
  • spring_workshops is the campaign name.
  • Organic Search and Email are default channel-group labels: broad, rule-based classifications.

A source / medium value combines origin and mechanism in one row label. A campaign dimension answers a different question: which named initiative was recorded.

Compatible acquisition dimensions

Choose dimensions from one scope family. The prefixes tell you whether the row concerns an original user acquisition or a particular session.

Use channel groups for a broad summary, then select a compatible more-specific dimension.
Question scopeBroad classificationOrigin and mechanismInitiative
First userFirst user default channel groupFirst user source / mediumFirst user campaign
SessionSession default channel groupSession source / mediumSession campaign

Practice: select the acquisition view

Choose the needed session dimension

Match each session-level question to the acquisition dimension that answers it.

Tap an item and then its matching pair.

Step 4 of 9

Interpret Engagement Without Mixing Units

Compare user and session units, calculate engagement rates, and interpret engagement time without treating it as proof of business success.

Read the unit before comparing

Users and sessions are different units

Before comparing engagement, identify the unit in each metric. A session is a visit context; one recognized person can start more than one session. An Active user is a distinct user who meets GA4 activity criteria, not a count of visits.

Read the exact user label in a report—such as Active users or Total users—before interpreting it. Do not automatically add user counts across channel rows: the same person may be represented in more than one row.

One user can create multiple sessions

A single visitor is shown having two separate sessions, illustrating that users and sessions are different counting units.

A recognized user may have multiple sessions; user totals are not visit totals.

Use compatible engagement measures

Define engagement measures precisely

For Harbor Workshops, a session is engaged when it lasts longer than 10 seconds, contains a key event, or includes at least two page or screen views.

Engagement rate = Engaged sessions ÷ Sessions. It is a proportion of sessions, so compare it with other rates only when the scope and period match.

Average engagement time per session measures focused or foreground activity, not simply elapsed visit duration. Its denominator is sessions. A per-active-user average would use a different denominator, so the two averages are not interchangeable.

What makes a session engaged

A diagram shows three alternative ways for a session to qualify as engaged and contrasts focused foreground activity with total elapsed visit time.

Any one qualifying condition can make a session engaged. Average engagement time counts focused or foreground activity, not all elapsed time.

Current-period practice extract

Current period: 01–14 Apr 2026 in UTC. These are selected channel rows, not property totals.

Harbor Workshops traffic-acquisition practice rows. `Key events` refers to `generate_lead`.
ChannelSessionsEngaged sessionsKey eventsSessions with generate_leadAverage engagement time per session
Organic Search1,000600403045 s
Email200140302060 s

Tip

Interpret carefully

Higher engagement can be useful evidence of audience activity, but it does not independently prove lead quality, revenue, or business success.

Apply the engagement definitions

Separate a user from sessions

A recognized visitor first arrives through Organic Search and later returns through the spring_workshops email campaign. How should this short timeline be counted?

Calculate the Organic Search rate

Organic Search recorded 600 Engaged sessions from 1,000 Sessions. Its engagement rate is ___.

Calculate the Email rate

Email recorded 140 Engaged sessions from 200 Sessions. Its engagement rate is ___.

Avoid an unsupported conclusion

True or false: A channel with longer Average engagement time per session automatically produces better business outcomes.

Step 5 of 9

Separate Key-Event Counts from Session Rates

Separate recorded generate_lead occurrences from the proportion of sessions that contain the selected key event.

Count the action and the session separately

Two related, different counts

Key events is an occurrence count for designated actions. In this case, it means recorded generate_lead occurrences, using the established Once per event counting method.

A session can contain more than one recorded occurrence. If one visitor submits two successful workshop inquiries during one session, that contributes 2 key events but only 1 session containing generate_lead.

Do not substitute an all-events count for a business-outcome count. All events can include routine interactions that are not workshop inquiries.

Occurrences versus qualifying sessions

Two session cards contain three generate_lead markers in total, while both cards qualify as sessions containing the event.

Repeated occurrences raise the key-event count without necessarily raising the number of qualifying sessions.

Read the two measures in one extract

The rate uses sessions, not occurrences

To compare an outcome across acquisition rows, use the same session scope for both parts of the rate:

Session key event rate = Sessions containing at least one generate_lead / Sessions

The numerator counts each qualifying session once, even if it contains repeated generate_lead occurrences. The denominator is all sessions in the same row and period.

Use an event-level extract when the question is about recorded occurrences. Use a compatible acquisition extract when the question is about session outcomes by channel. A property does not need to expose a standalone report named Key events for either question to be answerable.

Harbor Workshops: current-period extract

Selected channel rows for 01–14 Apr 2026 in UTC. Sessions with generate_lead is a supplied practice annotation, not a promised native GA4 report column.

These are selected channel rows, not property totals. `Key events` and the session annotation both refer specifically to `generate_lead`.
ChannelSessionsEngaged sessionsKey eventsSessions with generate_leadAverage engagement time per session
Organic Search1,000600403045 s
Email200140302060 s

Match the question to the unit

Choose the matching report view

Which extract is most appropriate for checking how many generate_lead occurrences were recorded during the current period?

Apply the session-rate definition

Calculate and justify the session rate

For Email, the extract records 30 generate_lead key-event occurrences, 20 sessions containing generate_lead, and 200 total sessions.

Calculate the session key event rate. Then explain in one sentence why 15.0% is not the session key event rate.

Write at least 60 characters (0/60).

Step 6 of 9

Make Like-for-Like Comparisons

Compare matched GA4 periods without confusing higher activity volume with a higher session outcome rate.

Check whether two extracts can be compared

Align the comparison context

A period comparison is defensible only when it describes like with like. Before interpreting a change, align:

  • Complete period length and weekday coverage
  • Property time zone
  • Web-stream scope, filters, and comparisons
  • Report dimension and its scope
  • Selected key event and metric definitions

For a recent period, also check that reporting is fresh enough to use. Do not compare a partial day or an incomplete period with a completed period.

A matched comparison can show what changed in recorded data. It does not, by itself, explain why it changed.

Match the reporting context

Two matched report cards show equal complete periods, UTC, the same filters, scope, dimension, and selected event; a partial period is excluded.

Match the reporting context before comparing results. Two dates alone do not make a comparison controlled.

Read the aligned extracts

Read the aligned extracts

Harbor Workshops provides two complete 14-day Traffic acquisition extracts. Both use the same web stream, UTC property time zone, no additional filters, the same session-scoped default channel-group dimension, and the selected key event generate_lead.

These are selected channel rows, not property totals. Sessions with generate_lead is a supplied practice annotation used to calculate the session key event rate; it is not assumed to be a native report column.

Current period

Current period: 01–14 Apr 2026 in UTC.

Traffic acquisition extract; selected channel rows only.
ChannelSessionsEngaged sessionsKey eventsSessions with generate_leadAverage engagement time per session
Organic Search1,000600403045 s
Email200140302060 s

Previous period

Previous period: 18–31 Mar 2026 in UTC.

Traffic acquisition extract; selected channel rows only.
ChannelSessionsEngaged sessionsKey eventsSessions with generate_leadAverage engagement time per session
Organic Search800480322445 s
Email160112201660 s

Calculate counts and rates separately

Volume growth is not rate improvement

Now separate a change in volume from a change in rate.

For Email, Key events rose from 20 to 30. The relative count change is:

(30 − 20) ÷ 20 = 50.0%

But the session key event rate uses sessions containing at least one generate_lead, not the Key events count:

  • Previous: 16 ÷ 160 = 10.0%
  • Current: 20 ÷ 200 = 10.0%

Email recorded more key-event occurrences while its session key event rate did not change. State a difference between rates in percentage points: 10.0% − 10.0% = 0.0 pp.

Compare counts and rates separately

A later-period event-count bar is taller than an earlier bar, while two session proportion rings have equal highlighted shares.

The Email count rises by 50.0%, while the share of sessions containing generate_lead remains 10.0% in both periods: a 0.0 pp rate change.

Tip

Keep the conclusion proportionate

The Email row has more Sessions as well as more Key events. More traffic can raise a count even when the share of sessions producing generate_lead is unchanged.

A careful finding is: “Email Key events increased 50.0%, while the Session key event rate remained 10.0%.” Do not turn that observation into an explanation of why the increase occurred.

Apply the comparison rule

Apply the like-for-like rule

Which comparison note is defensible for Email?

Step 7 of 9

State What Attribution Cannot Prove

Credit describes what was recorded

Reporting credit is not causal proof

Attribution assigns reporting credit to touchpoints that GA4 recorded. It can show that a channel is associated with a recorded key event, but it does not prove that the channel caused the outcome.

For example, a report cannot reveal whether a workshop inquiry would have happened without an email. Therefore, “Email caused additional inquiries” is stronger than the report evidence supports.

One journey can have several report views

A diagram shows a search visit, a later email-driven return visit, and a workshop inquiry, alongside separate report-view cards and dotted unseen paths.

For one recognized user, Google organic can describe the first arrival while newsletter / email describes a later session. Each report view summarizes a different recorded aspect of the journey.

Different views, different claims

Keep the reporting question and scope explicit

For Harbor Workshops, one recognized user first arrives through google / organic and later returns through the spring_workshops email campaign, shown as newsletter / email.

A first-user view can retain Google organic as the acquisition origin. A session-origin view can show Email for the later return session. An event-attribution view can describe reporting credit for touchpoints connected with generate_lead. These descriptions answer different questions; none alone proves which channel caused the inquiry.

Why a recorded journey is not the whole journey

Recorded journeys may be incomplete. A person might use another device, interact with an unrecorded channel, browse with restrictions, or decline analytics consent. These are visibility limits, not automatic proof of faulty tracking.

Because some touchpoints can be absent, do not give one channel all the credit or claim that it created incremental inquiries from a report alone.

Questions and defensible answers

Use the view that matches the question, then keep its limit visible.

Compatible report views can describe the same person without answering a causal question.
QuestionCompatible viewWhat it cannot establish
Where was this user first acquired?First user source / mediumWhich later session caused an inquiry
Where did this session originate?Session source / mediumWhich channel deserves all inquiry credit
Which recorded touchpoints received report credit?Event-attribution viewWhether an inquiry would occur without a channel

Test a causal claim

Check the strength of the claim

True or false: If a session acquisition result shows newsletter / email for a session with generate_lead, it proves that Email caused that inquiry.

Apply cautious attribution language

Rewrite the Email claim

A stakeholder writes: “Email caused 30 additional workshop inquiries in 01–14 Apr 2026.”

Rewrite this as a two-sentence evidence-based statement. First, state the recorded Email results using the supplied figures. Then name one visibility limitation relevant to the illustrated journey and explain what it prevents you from claiming.

Write at least 180 characters (0/180).

Step 8 of 9

Triage Report-Quality Warning Signs

Identify report-quality warning signs, frame them as hypotheses, and select evidence that can verify or reject them.

Treat anomalies as hypotheses

Signal, hypothesis, check

An unusual report pattern is a signal to investigate—not proof that tracking failed or that audience behavior changed. For example, fewer recorded generate_lead key events could reflect fewer submitted inquiries, an interruption in collection, or a reporting-context difference.

Use a proportionate sequence:

  1. Describe what the report shows.
  2. State a plausible hypothesis, not a diagnosis.
  3. Request evidence that could support or reject that hypothesis.

Do not inspect personal data or attempt technical changes yourself. Use available aggregate records, existing validation evidence, campaign documentation, and confirmation from an authorized property owner.

From pattern to verification

A flow from an unusual analytics pattern to several possible explanations and evidence sources.

Treat the report as the starting point for a check, rather than as proof of one cause.

Tip

Keep the conclusion conditional

A check can reject a hypothesis as well as support it. If aggregate successful-inquiry records are stable while recorded generate_lead key events suddenly fall, that mismatch is worth investigating—but it still does not identify the cause by itself.

Choose evidence that fits the signal

Read warning signs in context

Some labels and reporting treatments need careful reading. Their meaning depends on the report dimension and property context. Inspect any available data-quality indicator or notice before drawing a conclusion.

Warning signs are not diagnoses

Possible signals and proportionate evidence checks

Each pattern has more than one possible explanation.
Report patternPlausible readingUseful next evidence
Direct trafficNo referrer or campaign was recognized; not proof someone typed the address.Campaign-link register and referral context
UnassignedTraffic did not fit the applicable channel-group rules.Tagging records and property context
(not set)A value was unavailable for the selected dimension.Dimension scope and collection context
Unexpectedly low key eventsBehavior changed or collection may be incomplete.Aggregate inquiry record and existing validation log
Unexpectedly high key eventsMore actions occurred or occurrences may be duplicated.Existing validation log and owner confirmation
Very small denominatorA rate can be unstable.A larger compatible period or denominator

Warning

Withheld and modeled data

Thresholding can withhold data to protect privacy. Modeling can estimate otherwise unobserved activity where eligible. A withheld result is not automatically zero, and a modeled result is not proof of complete measurement.

For either case, inspect the available notice or indicator and relevant property context. Limit the strength of any performance conclusion until you know what the report is showing.

Practice: investigate without overclaiming

Triage the evidence

Match each report signal to the most proportionate next check. Treat every signal as a hypothesis, not a confirmed tracking fault.

Tap an item and then its matching pair.

Step 9 of 9

Write an Evidence-Based Performance Note

Produce a short, defensible interpretation of supplied acquisition, engagement, and key-event results.

Use four distinct sections

Build a note that can be checked

A defensible performance note makes four different jobs visible:

  • Observed — figures that the supplied, compatible reports show.
  • Interpretation — a cautious meaning of the pattern.
  • Limits — what attribution and measurement cannot establish.
  • Next check — one proportionate action to verify an important uncertainty.

Keep facts separate from explanations. A channel row can show an association with sessions or key events; it does not prove that the channel caused the outcome. Do not turn one report comparison into a spending or optimization recommendation.

A four-part reporting-note structure: observed figures, cautious interpretation, limits, and a next check.

Use this sequence to separate what the report records from what still needs verification.

Tip

Keep the claim proportionate

For Harbor Workshops, an appropriate limitation is that recorded attribution does not prove incremental impact. One recognized user may first arrive through Google organic search and later return through an email campaign. A useful next check can inspect an existing aggregate inquiry record or an existing validation log; it must not claim that tracking is broken before evidence supports that conclusion.

Use compatible evidence

Read the supplied session-origin extract

The question for the final note is about where sessions originated. Use Traffic acquisition with the session-scoped dimension Session source / medium.

The tables below compare complete, compatible 14-day periods in UTC, with the same web-stream scope and no additional filters. They are selected channel rows, not property totals.

From these rows, Organic Search has a 60.0% engagement rate and 45 s Average engagement time per session in both periods. Email has a 70.0% engagement rate and 60 s in both periods. Email Key events rise from 20 to 30, but its sessions also rise from 160 to 200. Its session key-event rate is unchanged: 16 / 160 = 10.0% and 20 / 200 = 10.0%.

Current period

Current period: 01–14 Apr 2026. Selected channel rows only.

Authoritative practice extract; generate_lead is the selected key event.
Channel — source / mediumSessionsEngaged sessionsKey eventsSessions with generate_leadAverage engagement time per session
Organic Search — google / organic1,000600403045 s
Email — newsletter / email200140302060 s

Previous period

Previous period: 18–31 Mar 2026. Selected channel rows only.

The same metric definitions, scope, and column order apply.
Channel — source / mediumSessionsEngaged sessionsKey eventsSessions with generate_leadAverage engagement time per session
Organic Search — google / organic800480322445 s
Email — newsletter / email160112201660 s

Final application

Write the evidence-based performance note

Harbor Workshops asks: “Which origins brought sessions, and what can we responsibly say about Email performance?”

Write a short note comparing 01–14 Apr 2026 with 18–31 Mar 2026. Begin with the report and scoped dimension you selected. Then use exactly these headings:

Observed
Interpretation
Limits
Next check

Your note must cite an accurate engagement comparison and accurate Email key-event and session-rate figures. Distinguish the higher Email key-event count from its unchanged session key-event rate. State one attribution or measurement limit, then propose one justified verification check. Do not claim that a tracking problem is proven.

Write at least 350 characters (0/350).

Close with evidence, not certainty

Summary

Performance-note checklist

A short note is stronger when another person can trace each statement back to a compatible report row or to a clearly named uncertainty.

  • Use Traffic acquisition and Session source / medium for a question about where sessions originated.
  • State observed counts, rates, and engagement measures without treating them as causes.
  • A higher Key events count can occur while the session key-event rate is unchanged.
  • Label attribution and measurement limits explicitly.
  • End with a targeted evidence check, not an unsupported optimization or spending decision.

Step complete

Congratulations! You completed: Read GA4 Acquisition, Engagement, and Key-Event Reports

You can now turn compatible GA4 acquisition, engagement, and key-event evidence into a concise note that distinguishes observed results, limits, and a justified next check.

Download the App now to have access to + 5000 free courses, exercises, certificates and lots of content without paying anything!

  • 100% free online courses from start to finish

    Thousands of online courses in video, ebooks and audiobooks.

  • More than 60 thousand free exercises

    To test your knowledge during online courses

  • Valid free Digital Certificate with QR Code

    Generated directly from your cell phone's photo gallery and sent to your email

Cursa app on the ebook screen, the video course screen and the course exercises screen, plus the course completion certificate