# GA4 key events vs conversions, and what to import into Google Ads

Canonical URL: https://www.martechdevs.com/blog/ga4-key-events

By [Jafrin Akter Alo](https://www.martechdevs.com/authors/jafrin-akter-alo)

Data Engineer and Co-founder, martechdevs

Published: 2025-04-08

Last updated: 2026-09-02

A GA4 key event marks an action that matters to your business. A Google Ads conversion makes an outcome available for advertising measurement and, when the campaign's goal configuration uses it, bidding. The decision is not simply which events to import. Choose the business outcome, define how it is counted, verify its collection and then select the measurement source that supports the campaign.

## Separate the action, its business importance and its use in advertising

An event records an action, such as a registration or purchase. Marking it as a key event gives that action significance in Analytics reporting. Creating an Ads conversion from it is another configuration step. Google's [key events and conversions explanation](https://support.google.com/analytics/answer/13965727?hl=en) describes that relationship.

Keep the three decisions explicit:

- **What happened?** The application confirmed a new workspace trial.
- **Why does it matter?** The trial is a measurable step toward becoming a paying customer.
- **Should a campaign optimise toward it?** That depends on the trial's quality, measurement reliability and relationship to the campaign's actual objective.

A tag firing successfully does not establish that the event is a useful bidding target. A high-volume event can be easy to measure and still encourage the wrong acquisition behaviour.

![Business outcome moves through event validation, GA4 key-event marking, conversion-source selection, counting checks and campaign goal review](https://www.martechdevs.com/assets/blog/ga4-key-events.svg)

The GA4-based route is one option in that design. A native website conversion or an offline business outcome can also provide an advertising signal. Choose from the outcome backwards.

## Mark events that represent a meaningful result

For a self-serve SaaS, useful candidates might include a confirmed new trial, a defined activation milestone or the first successful payment. For a sales-led business, they might include an accepted demo request, a qualified lead or a converted customer.

Write a one-sentence contract before marking the event. For example: “`trial_started` occurs once when a previously ineligible workspace receives its first active trial entitlement.” That excludes clicking the trial button, retrying a failed request and adding another user to an existing workspace.

Document five details:

1. The entity being counted: person, workspace, lead, account or transaction.
2. The evidence of success: application state, accepted form response, CRM status or payment record.
3. The conditions that should not trigger the event.
4. The repeat and retry rules.
5. The owner who can verify the outcome against business records.

Use your [tracking plan](https://www.martechdevs.com/blog/tracking-plan) to hold this contract. An event-name list without trigger definitions leaves room for every integration to count a different version of the outcome.

Do not mark every interaction merely to make the key-event rate look stronger. A scroll, a pricing-page visit and an accepted lead request can all provide information, but they do not represent equal commercial progress. Keep diagnostic interactions available for analysis without automatically promoting them into acquisition goals.

Review the actual key-event list, including automatic markings. Google documents defaults and additional markings associated with linked advertising features in its [key-event setup guidance](https://support.google.com/analytics/answer/13128484?hl=en). The list may contain more than the events your team consciously selected.

## Understand the three different counts

Event count, key-event count and key-event rate are not interchangeable. The first counts collected events. The second follows the key event's counting configuration. A rate asks how many sessions or users contained the selected outcome, relative to the relevant population.

GA4 supports counting a key event once per event or once per session. A counting-method change affects future key events, not past data. [Google's counting-method documentation](https://support.google.com/analytics/answer/13366706?hl=en) explains those options.

For a purchase event, two real transactions in a session are two different business outcomes. For a registration event, repeated firing on refresh may be a bug. Changing to once per session can conceal that bug without preventing it from affecting other destinations or later sessions.

### Work through a small example

Consider an illustrative, fully observed dataset with 10 users and 12 sessions. Only one event type is selected as the key event. User A triggers it twice in one session and once in another. User B triggers it twice in a third session. Nobody else triggers it.

- Event count for that event type is 5.
- Key-event count with once-per-event counting is 5.
- Key-event count with once-per-session counting is 3.
- Sessions containing the selected key event total 3, so session key-event rate is 3 ÷ 12 = 25%.
- Users containing the selected key event total 2, so user key-event rate is 2 ÷ 10 = 20%.

Google defines the session rate as sessions with a key event divided by total sessions in the [Traffic acquisition report](https://support.google.com/analytics/answer/12923437?hl=en). It defines the user rate as users performing a key event divided by total users in the [User acquisition report](https://support.google.com/analytics/answer/12922540?hl=en).

The five occurrences do not imply a 41.7% session conversion rate. Five divided by twelve is an event frequency per session, not the share of sessions containing the outcome. Keep the selected event, identity and report scope consistent when comparing rates.

Also distinguish GA4's session counting from the Google Ads Count setting. Ads offers One or Every for a conversion action after an ad interaction; One is not a permanent deduplication rule for a person, company or CRM lead. [Google Ads counting options](https://support.google.com/google-ads/answer/3438531?hl=en) describe that separate basis. Validate the actual action and attribution configuration rather than assuming the two products' settings are synonyms.

## When importing a GA4 key event is a sensible choice

Use the GA4 route when you already have a reliable, well-defined event and its conversion capabilities fit the advertising requirement. This can keep the business definition consistent between acquisition analysis and the signal shared with Ads.

Before creating the conversion, verify the linked property and Ads account, auto-tagging and access. Google supports creating these conversions through its Analytics and Ads interfaces. Native Google Ads measurement is required for view-through conversions; imported Analytics conversions do not provide that feature. [Google's import guidance](https://support.google.com/google-ads/answer/10632359?hl=en) describes the requirements and limitation.

Importing does not repair an event that fires before success, a broken consent sequence or a registration definition that combines prospects with invited teammates. Check the underlying collection path with the [GA4 SaaS setup guide](https://www.martechdevs.com/blog/ga4-setup-saas) first.

A practical acceptance record should identify the source property and stream, event name, business definition, counting method and conversion action created. Record when the new configuration became effective so a later report does not silently compare two different implementations.

## When a native website conversion is the better fit

Choose a native Google Ads website conversion when its measurement features or operational ownership better fit the campaign, including a requirement for native view-through reporting. That is a feature-based decision, not proof that Analytics imports are inherently low quality.

Keep the business trigger aligned with the Analytics event if both are measuring the same outcome. Otherwise a native Ads purchase on a confirmation-page view and a GA4 purchase on confirmed payment may disagree because they are different definitions.

Use one documented primary measurement path per equivalent outcome in the campaign. A second action can be useful for a controlled comparison, but do not simply include both in bidding and assume Google will recognise them as the same sale.

There is a current, specific exception to a simplistic “one data source” rule. Google's **multi-source conversions beta** can supplement an eligible code-based website conversion with backend data. Matching uses `transaction_id` within the same action; it does not deduplicate across separate conversion actions. The documented feature does not support imported Analytics or URL-based actions, and uploaded matched values can overwrite tag values. [Google's beta documentation](https://support.google.com/google-ads/answer/16542291?hl=en) describes those boundaries.

If you adopt that feature, treat it as a deliberate implementation with value, identifier and consent checks. The additional source has a 14-day trial before becoming biddable automatically. Do not mistake the trial for an indefinite observation-only setting.

Matched value updates are disabled during that initial trial. Google's [Data Manager event-processing guide](https://developers.google.com/data-manager/api/devguides/events/google-ads/offline/send-events) says uploaded values only override the tag's reported value after the trial ends. Check the transition date before treating an unchanged trial-period value as an upload failure.

## When the right signal is an offline business outcome

If the outcome happens in a CRM or billing workflow, measure that outcome where it becomes known. A form submission cannot tell you whether sales accepted the lead or whether the customer paid an invoice later.

Define the stage precisely and preserve its real timestamp. For example, “qualified lead” might require the right market, a relevant use case and an accepted sales handoff. Merely opening the CRM record should not create the event.

Choose the supported upload mechanism for the current integration. Google's [2026 enhanced-conversions update](https://support.google.com/google-ads/answer/16884284?hl=en) describes unified settings and the migration of offline conversion and enhanced-conversions-for-leads uploads toward the Data Manager API, with legacy access qualifications. An old Google Ads API tutorial is not sufficient evidence that a new integration can use that path today.

Keep an upload ledger: business record, outcome, occurred-at time, permitted matching identifiers, destination action and processing result. An upload accepted for processing is one stage; an attributed conversion available in the intended report is another.

Do not invent missing click IDs, backdate outcomes or replace a failed qualification with a softer event to make the upload succeed. If coverage is limited, quantify the eligible population and the unmatched share so the campaign owner understands the signal.

## Decide what each campaign should use for bidding

Primary and secondary describe a conversion action's optimisation role, not its importance to every team. Google says primary actions are used for bidding when their standard goal is used by the campaign. Secondary actions are usually observational, but actions in a custom goal can be used for bidding regardless of that label. [The primary and secondary reference](https://support.google.com/google-ads/answer/11461796?hl=en) explains this exception.

Inspect account-default goals and any campaign-specific overrides. Do not assume an action marked primary is necessarily the goal a particular campaign uses, or that secondary guarantees exclusion from every custom goal.

Select the closest reliable signal to the campaign's economic objective that the team can measure with useful frequency and delay. There is a tradeoff: a late closed-won event may express value well but arrive too slowly or sparsely for the intended operating cadence. A trial is earlier, but can reward poor-quality acquisition if it has a weak relationship with paying customers.

Check that relationship using mature cohorts. Compare the fraction of trials that become qualified or paid, broken down by meaningful acquisition groups. A larger trial count accompanied by falling paid conversion is a reason to inspect the goal and audience quality, not immediately celebrate cheaper acquisition.

### Avoid adding the same customer's value at every stage

In an illustrative mature cohort, 100 leads produce 20 qualified leads and 5 wins. Assume each win contributes USD 2,000 of the same defined gross-profit measure. The cohort contributes USD 10,000, implying USD 100 per initial lead or USD 500 per qualified lead under those assumptions.

Those are alternative expected values at different stages. They are not three independent revenue streams. Adding lead value, qualified-lead value and full customer value into one undifferentiated optimisation total can count the same economic opportunity repeatedly.

Choose a stage/value design deliberately. State whether values are observed revenue, estimated future value or an incremental stage value. Recalculate estimates on mature cohorts and document the uncertainty. Do not assign a high arbitrary value to the event you want the dashboard to favour.

## Reconcile reports before changing the implementation

When GA4 and Ads differ, compare the same event or conversion action, date basis, time zone, counting rules, attribution scope and processing state. Distinguish all-channel business outcomes from outcomes attributed to advertising.

Google Ads commonly reports attributed conversions against the ad interaction date. Its conversion-time columns support a different comparison. The [offline discrepancy guide](https://support.google.com/google-ads/answer/13321563?hl=en) recommends the relevant “by conv. time” columns and warns against comparing upload dates with conversion reporting.

If an ad interaction happens near the end of one month and the outcome happens in the next, the same underlying conversion can appear in different monthly comparisons. Changing a tag will not resolve a date-definition mismatch.

Separate three questions in the review: did the event occur, did the destination receive the intended record, and how did the reporting model credit it? Follow the [conversion diagnostics guide](https://www.martechdevs.com/blog/google-ads-conversion-diagnostics) when the collection or destination stage is failing.

## Use a release checklist for goal changes

- [ ] The outcome has a success condition, unit, owner and exclusion rules.
- [ ] Failure, refresh and retry tests do not create unintended duplicates.
- [ ] GA4 counting and Ads counting are documented separately.
- [ ] The chosen source supports the required advertising features.
- [ ] Equivalent actions are not unintentionally included twice in campaign goals.
- [ ] Primary, secondary, custom goals and campaign overrides are inspected.
- [ ] Values have a defined unit and do not silently add the same opportunity across stages.
- [ ] The change date and downstream quality checks are recorded.

If the team needs help reconciling tags, CRM outcomes and campaign goals, an [analytics and tracking review](https://www.martechdevs.com/analytics-tracking) should deliver an outcome contract and evidence that the chosen signal reaches the intended action. A green status alone is not the commercial acceptance criterion.

## Frequently asked questions

### Are GA4 key events just the old conversions renamed?

The terminology changed, but the operational distinction now matters: Analytics key events identify business-important actions, while advertising conversions have their own creation and goal configuration. Review the existing settings rather than assuming a renamed label proves your campaigns use the intended outcome.

### Should I import all my GA4 key events into Google Ads?

No. Import the outcomes that support a defined advertising measurement or optimisation purpose. Keep useful product and journey metrics in Analytics even when they do not belong in campaign bidding.

### Is a native Ads tag always better than a GA4 import?

No. Choose based on required capabilities, collection reliability and ownership. Native measurement supports features such as view-through reporting that the Analytics import does not. A correctly defined imported event can still be an appropriate signal for another requirement.

### Why is the key-event rate different from key events divided by sessions?

The rate measures sessions containing the selected outcome. One session can contain several occurrences, so total occurrences divided by sessions measures something else. Use the metric whose denominator matches the question.

### Can I use once-per-session counting to fix duplicate events?

It can reduce the displayed key-event count without repairing the event stream. Find the duplicate trigger or retry behaviour, preserve genuine repeat transactions and verify every destination affected by the implementation.

## Sources

1. [Google: Conversions versus key events](https://support.google.com/analytics/answer/13965727?hl=en)

2. [Google: Mark events as key events](https://support.google.com/analytics/answer/13128484?hl=en)

3. [Google: Change key-event counting](https://support.google.com/analytics/answer/13366706?hl=en)

4. [Google: Traffic acquisition report](https://support.google.com/analytics/answer/12923437?hl=en)

5. [Google: User acquisition report](https://support.google.com/analytics/answer/12922540?hl=en)

6. [Google Ads: Conversion counting options](https://support.google.com/google-ads/answer/3438531?hl=en)

7. [Google Ads: Create conversions from Analytics key events](https://support.google.com/google-ads/answer/10632359?hl=en)

8. [Google Ads: Multi-source conversions beta](https://support.google.com/google-ads/answer/16542291?hl=en)

9. [Google Ads: Enhanced-conversions settings updates](https://support.google.com/google-ads/answer/16884284?hl=en)

10. [Google Ads: Primary and secondary conversion actions](https://support.google.com/google-ads/answer/11461796?hl=en)

11. [Google Ads: Offline conversion discrepancies](https://support.google.com/google-ads/answer/13321563?hl=en)

12. [Google Data Manager: Multi-source event processing](https://developers.google.com/data-manager/api/devguides/events/google-ads/offline/send-events)