Comparing metrics: Google Analytics 4 vs. Universal Analytics

Counting method: UA counts one goal per session, while GA4 lets you choose between once per event or once per session.

For example, when a user completes a goal 5 times in a single session, UA will show one conversion while GA4 will show 1 or 5 key events, depending on the counting method you choose.

Note: Once per event is the default counting method for most key events, except those created in an automatically created GA4 property or using the Setup Assistant goals migration tool.

The default settings can cause a higher key event count in GA4.

Update the key event counting settings. To align GA4 counting to UA goals counting, make sure the GA4 key event counting method is set to Once per session.

Notes:

• The GA4 UI will recognize that this isn't a recommended setting as it can lead to the under counting key events. For example, if you're a lead-generation business, you want to count every lead submission, not just one per user.
• Google Tag Manager has controls over how and when tags can fire. Learn more about Tag firing options.

This recommendation isn't relevant when comparing UA ecommerce transactions because they're already counted as Once per event in UA.

Site coverage: If the UA tag is implemented across a different set of pages within a website than the GA4 tag, measurement gaps can emerge. Especially if a landing page from a Google Ads ad isn't accurately tagged, it could be that the information you need to accurately measure and attribute a key event to that ad interaction isn't available.

Depending on whether GA4 tags are implemented on more or fewer pages than UA tags, GA4 will respectively show more or fewer users, sessions, page views, and subsequent key events.

We recommend relying on a standardized method of implementation to make sure tags are implemented across your site (for example, the Google tag or Google Tag Manager). Also consider settings you may have configured in the past, such as cross-domain measurement.

Use the Tag coverage summary to confirm if the Google tag is implemented across your website. Learn more About the Tag coverage summary.

To minimize discrepancies, rely on the same tag implementation methods for UA and GA4.

Implementation errors: When setting up data collection for GA4, it's possible a tag is implemented incorrectly. This can result in noticing no data at all in your GA4 property, but in some instances, you’ll notice partial data in your GA4 property. If so, you’ll notice discrepancies when comparing it to UA.

It's likely that a broken implementation shows lower key event counts for GA4 than UA.

We recommend you rely on standardized implementation methods outlined in the Developers Guide instead of non-standard implementations. A tag management system can help bring consistency in implementations.

Filters: UA filters work very differently from GA4 filters. It's common for UA filters to be in place and significantly alter the data inside a UA property, for example, "only show data from France".GA4 has a very different set of possible filters.

If UA filters reduce reported traffic, this can result in GA4 showing higher amounts of traffic.

Data filters: Create include/exclude filters for internal and developer traffic.

Event modifications and custom events: Modify event names and parameters.

Identify unwanted referrals: Include only the referrals you want.

Create subproperties (360 only): Create fully-functional properties that are subsets of the data in your 360 properties.

Referral exclusions: Exclusions set in UA can affect key events attributed to Google Ads. If these exclusions are not set up accordingly in GA4, the credit attributed to Google Ads can differ, for example, exclusions for payment providers like PayPal are often made in UA.

If referrers were excluded in UA but not in GA4, GA4 key events can be misattributed in GA4, resulting in fewer key events attributed to the Google Paid channel when compared to UA. Referral exclusions affect the amount of key event credit.

Match referral exclusions settings between UA and GA4 to minimize incorrect attribution. Note that excluding referrals does not change the overall amount of key events in the Google Analytics property it only affects how key events are attributed.

Learn more about Referral exclusions in UA and how to Identify unwanted referrals in GA4.

Key event lookback window: Key event lookback window settings in Google Ads define how long after a Google Ads touchpoint credit can be attributed to that touchpoint. Often this setting is set to a value of choice, for example, 90 days.

A difference in settings can mean a touchpoint gets attributed key event credit by UA while it doesn't by GA4, or vice versa.

Attribution model settings (in Google Ads): Attribution models affect how credit is distributed across touchpoints in a conversion path. Differences in attribution model settings in Google Ads between the UA and GA4 can result in different credit allocations across campaigns in a Google Ads account or manager account (MCC).

Note: Changing attribution settings in Google Ads does not affect the overall amount of credited attributed to Google Ads.

Reporting methodologies in Google Ads: In Google Ads, to show which ad drove a key event, key events are reported based on ad interaction time reporting. In Google Analytics, key events are reported differently, using key event time reporting.

For example, if a key event that took place on May 10th, it can be attributed to an ad click on May 5th. Google Analytics generally reports this key event on May 10th, while Google Ads will report this key event on May 5th.

This effect applies both to UA and to GA4 imported events in Google Ads, although differences in settings (like attribution model or key event lookback window) between key event actions can amplify it.

If you align settings across key event actions imported into Google Ads, you minimize the effects of reporting methodologies.

Keep in mind that key events can be attributed during the full length of the key event lookback window, up to 90 days in Google Ads, so it can take up to 90 days for GA4 and UA to fully compare to one another.

To assess whether a key event pair can be compared, or whether you must wait longer, use the Path metrics report in Google Ads for the UA key event to determine how long it takes the majority of users to interact with a key event on the path to a key event.

Note: More than 95% of key events get attributed within the first 14 days. This means you must use Google Ads data in your comparison between UA and GA4 that is at least 14 days old.

Ecommerce measurement: Google Analytics is compatible with various ecommerce schemas. You can implement the GA4 schema and the UA schema simultaneously, or rely on one schema for both properties. If you use two schemas, data discrepancies can emerge. Review the Developer Guide for more information.

Differences in ecommerce schema can cause key event counts in GA4 to be higher or lower than in UA.

When collecting ecommerce data for GA4, it's best practice to rely on the GA4 ecommerce schema (Developer Guide).

It's recommended that you not rely on connected site tags if you’re using ecommerce measurement.

It's also recommended you use the same ecommerce schema for UA and GA4 when comparing the two.

Tag firing conditions: Hard-coded tag firing conditions can affect how and when a tag fires.

Since setting up GA4 data collection in many cases consists of implementing new tags, it's possible previous firing conditions need to be implemented for GA4 for them to take effect. We recommend firing tags according to the same conditions to minimize data collection differences.

If data collection is restricted in UA and not in GA4, then user, session, page view, and key event counts can be higher in GA4.

It's recommended you rely on gtag.js or gtm.js (using GA4 tags) to collect data for GA4 properties.

Tag firing conditions are not directly impacted if you use other forms of tagging (for example, connected site tags), but retagging is the first recommendation in case tag firing conditions cause tags to fire inconsistently across UA and GA4.

Exclude internal traffic: Settings can be used to filter out internal users (from a customer’s perspective, i.e., a customer’s employees) or test traffic.

If settings differ in UA and GA4, the property without the exclusion filters is expected to show higher user, session, page views, and subsequent key events.

Page changes based on browser history events: GA4’s enhanced measurement by default measures page changes based on browser history events; UA does not. This will cause higher pageviews in GA4 than UA and can lead to discrepancies.

For example, if your site’s construction and usage patterns cause many pages to be loaded via changes in the browser history, enabling this setting will create measurement for these events. This will result in discrepancies from UA where these events haven't been measured.

Use matching setups for similar UA conversions and GA4 key events in Google Analytics.

You can configure newly created events from these sources to distinct event names so that UA and GA4 measurement can match more closely. If it isn't possible to set up in the same exact way, the two should not be compared.

Spam and bot filters:

• Spam: In UA, users must manually set up spam filters using a view filter, but GA4 works differently.
• Bots: In UA, users can disable bot filters in the view settings, they’re on by default. In GA4, bots are automatically filtered.

Filters reduce traffic in UA. If not applied in GA4, GA4 will report more traffic and subsequent key events, in cases where spam/bot traffic manages to interact with a key event on the path to a key event.

Confirm if URL parameter exclusions are used in UA to understand potential differences in GA4.

DDA model for key events and revenue: In UA, the attribution model is last non-direct click for goals. GA4 key events use the data-driven attribution (DDA) model by default, but are modifiable.

Depending on how credit shifts based on DDA, users can notice more or less credit being attributed to paid (Google) channels. Note that the attribution model doesn't affect the overall amount of key events for a given event.

Although not recommended, this can be tested/quantified by switching the property’s attribution model to Last click in GA4.

Time zone differences and campaign reset: UA cuts a session off at midnight and restarts a session with new campaign parameters; GA4 does not. This can lead to session discrepancies, which can then lead to goals/session key event discrepancies.

This isn't expected to drive any significant differences in the total amount of key events in a given property. It's expected to only be visible if individual days are compared in UA vs GA4.

GA4 can be configured to measure pageviews based on browser history events; UA doesn't do this. This leads to richer measurement in GA4 but can be a source of discrepancies. You can disable this setting if you want the numbers to more closely align between UA and GA4.

Google signals and User ID: These settings help deduplicate users in GA4 properties. In UA, both settings only affect a few reports and a separate view. In GA4, they affect all information in the property.

Turning on Google signals and User ID in GA4 reduces new/overall user counts, connecting interactions that in UA would be considered from multiple users. As such, credit can be attributed differently. These features aren't expected to affect the overall amount of conversions within Google Analytics but can result in differences in exports to Google Ads. If an advertiser measures engaged-view conversions (EVCs), they can be influenced by the use of Google Signals and User ID.

No direct solution – GA4 deduplication provides better insight into how users interact with a customer’s assets.

We recommend implementing both features in both platforms to minimize discrepancies, although they can't be prevented.

Cookie override

When you change cookie settings in UA, you need to match the setting when you configure GA4.

Reducing cookie expiration dates can affect attribution and result in differences in key event attribution. Learn more about configuring your Google tag settings.

For connected site tags (including sideloading) only: Connected tags load sequentially; that is, the GA4 tag loads after the parent tag (an existing gtag.js or analytics.js tag). This can mean:

1. Loading multiple tags can have an impact on website speed.
2. If users navigate extremely quickly through a website, tags may not have finished rendering, resulting in missed events.

Key event modeling from consent mode: Key event modeling is available in Google Ads for advertisers who have create Google Ads conversions based on Analytics key events from properties implemented using gtag.js or Google Tag Manager.

For analytics.js implementations, we don't offer modeling in Google Ads for conversions. This is because the consent state can't be defined accurately without having a gtag.js or Google Tag Manager implementation.

If both UA and GA4 rely on tags compatible with consent mode, both can receive key event modeling in Google Ads.

In GA4, this shouldn't lead to an increase in key events for web traffic, but it’s possible that more credit is attributed to the Google Ads channel in GA4 than in UA, leading to better Google Ads performance.

To compare this, you can segment campaign – or account level – reporting in Google Ads. Make sure to compare key event actions that measure the exact same key event action.

Credit export model: The model by which key events are exported from GA4 to Google Ads is different from UA. Where UA, and initially GA4, relied on a last non-direct click export model, GA4 now uses a final credit export model, allowing for fractional credit to be exported, even if the last non-direct touchpoint was not a Google Ads touchpoint.

Although this isn't resulting in property-level differences, it can result in a different number of key events exported to Google Ads. This number can be both higher or lower when compared to UA and is dependent on a user’s independent user journeys.

It's recommended to align key event counting settings for UA and GA4 key event pairs. Learn more About conversion counting options.

Applies to manually created GA4 properties and Smart Campaign customers: The GA4 Google paid channels export model attributes all key events exported to Google Ads to Google channels.

This results in more key events being exported to Google Ads than the Paid and organic channels setting because it excludes non-Google paid channels from the key event path.

Using Google paid channels can provide deeper insight into how paid Google interactions drive impact. This model isn't available in Universal Analytics, so it can be difficult to compare similar UA and GA4 events when imported to Google Ads.

By default, new GA4 properties are set to Google paid channels.

You can use the GA4 property-level attribution settings in the Admin section to control how key events are exported to Google Ads.

In order for GA4 to factor in non paid channels when exporting conversions, use the Paid and organic channels setting. Learn more about selecting attribution settings.