We wrote previously about how importing Universal Analytics (GA3) conversions into Google Ads is a half-baked solution. Google Ads only sees those GA3 conversions per the last non-direct click attribution model that GA3 applies by default. Google Analytics 4 (GA4) is a significant departure from GA3. Let’s explore how GA4 imported conversions are attributed in Google Ads.
GA4 collects several events automatically, including the session_start event, which designates the start of a session. GA4 events are associated with a session in two ways: (i) they have the same ga_session_id, and (ii) they have the same ga_session_number. It is important to note that with GA4 a new campaign does not produce a new session_start event unless the GA4 web session timeout has expired, whereas GA3 starts a new session with each new campaign regardless of its session timeout settings. Also, GA4 hits can arrive up to 72 hours after they happen (so report data in the same historic date range can change over time).
How does GA4 handle attribution?
GA4 attribution settings can be found in the GA4 admin at “Property” > “Attribution Settings”. I was hoping imported GA4 goals would expose more ad attributed conversions to Google Ads (i.e., more like the click attribution of Google Ads website conversions that send data from the browser directly to Google Ads), but some new documentation does not support that notion.
This GA4 help content, “[GA4] About attribution and attribution modeling“, inventories GA4 admin “Attribution Settings”:
- Reporting attribution model.
- Lookback window.
Important takeaways:
- Cross-channel last click is the only last click model that you can export to Google Ads.
- Ads-preferred last click is only available for reporting purposes.
- Data-driven attribution model is not available yet, but will be coming soon.
- Changing the reporting attribution model in these settings does not impact attribution models in the GA4 Advertising > Attribution > Model Comparison report. But, such changes will be otherwise reflected in GA4 reports with conversion and revenue data, and applies to both historical and future data.
- These changes will specifically be reflected in all reports that use event-scoped traffic dimensions, for example: Source, Medium, Campaign, and Default channel grouping.
- User and session scoped traffic dimensions–such as Session source or First user medium–are unaffected by changes to the reporting attribution model.
- The User acquisition report uses first-touch attribution.
- The Traffic acquisition report uses last-touch attribution.
- All attribution models exclude direct visits from receiving attribution credit, unless the path to conversion consists entirely of direct visit(s).
- Changing the lookback window only applies going forward, and impacts all GA4 reports.
- Attributed conversion data can still be updated for up to 7 days after the conversion is recorded. In analysis, consider selecting a date range before or after the previous week.
From the snag below of example scenarios, it seems that GA4 imported goals will be treated just like GA3 imported goals in Google Ads conversion reporting.
Google Ads imported GA4 conversions reporting
Imported GA4 conversions are reported in Google Ads as conversions if the related user action:
- Is associated with a Google Ads ad interaction for the GA4 linked ad account, per the GA4 cross-channel last click attribution model exported to Google Ads.
- Happens within configured conversion windows for the respective conversion action.
- Happens in the reported date range (considering conversion date vs. ad interaction date).
Google Ads allows controlling several attributes of an imported GA4 conversion: value, count (every conversion, or just one), click-through conversion window (7 to 90 days), and attribution model. The GA4 admin settings lookback window does not impact this imported data.
Comparing GA3 imported vs. GA4 imported vs. Website Google Ads conversions
Consider the following 5 interactions. Assume they are from the same user on the same device, with a 6 month session timeout in both the GA3 and GA4 properties tracking this activity, and continuous tracking integrity (i.e., tracking cookies not deleted or suppressed).
- DAY 1 = Paid Google Ads ad click (conversion ✔)
- DAY 5 = Organic search result click
- DAY 17 = Direct visit (conversion ✔)
- DAY 29 = Paid Google Ads ad view
- DAY 32 = Email campaign link click (conversion ✔)
GA3 imported conversions (with GA3’s last non-direct click attribution model inherently applied).
- Google Ads conversion settings: 30 day click-through conversion window; counting every conversion.
- Conversions attributed:
- (1) Paid Google Ads ad click (conversion ✔)
GA4 imported conversion (with GA4’s cross-channel last click attribution model applied).
- Google Ads conversion settings: 30 day click-through conversion window; counting every conversion.
- Conversions attributed:
- (1) Paid Google Ads ad click (conversion ✔)
Google Ads Website conversion (where Google Ads account tracking code sends the conversion signal from your website to Google Ads).
- Google Ads conversion settings: 30 day click-through conversion window; 7 day view-through conversion window; counting every conversion.
- Conversions attributed:
- (1) Paid Google Ads ad click (conversion ✔)
- (3) Direct visit (conversion ✔)
- (5) Email campaign link click (conversion ✔)
Google Ads website conversion’s 7 day view-through conversion window allows it to pick up an additional attributed conversion, and not being pinned to the last non-direct click allows it to pick up one more additional attributed conversion.
Modeled conversions to address signal loss
As of July 2021, some of the conversions reported in GA4 can include both “modeled conversions” and actual observed conversions. Such modeling is one of the ways Google is combating signal loss. These modeled conversion make their way into Google Ads when you import GA4 conversions. However, Google does not differentiate between observed and modeled conversions in reporting.
When looking at GA4 reports, keep in mind that attributed conversion data for each channel can still be updated for up to 7 days after the conversion is recorded. This is because Analytics is processing that data and using it for model training. GA4 imported conversions as reported in Google Ads can experience the same phenomenon.
First party data to train conversion models
The more actual observed conversions sent to Google, the better the model training, the higher quality the modeled conversions. Model training is also improved the more first party data you associate with conversion data sent to Google (as always, with consideration for user consent). Google Signals is the identity software that uses Google accounts to recognize logged-in users. GA4 uses this feature to recognize users instead of relying on first-party cookies. Google also uses first party data available from new collection techniques like enhanced conversions, and store sales measurement for retailers and restaurants (or other business types).
NOTE: In GA4 if there isn’t enough traffic to inform the model, modeled conversions are attributed to the “Direct” channel.
Server-side GTM to address signal loss
The Google Ads Website conversion leans upon third party cookies, meaning stringent default browser security settings (like Safari ITP) disintegrate such reporting by truncating the lifespan for JavaScript set cookies. If you can’t see a return visitor as “returning” more than 7 days past their initial browser visit (and in some scenarios for only 24 hours), that is very disruptive to conversion tracking for any analytics or advertising platform, impacting both click-through and view-through conversion attribution.
Server-side GTM can also help help defeat such signal loss (as always, with consideration for user consent), and just as importantly allows advertisers to regain control of tagging & data related to their advertising. If you put a vendor’s tag on your website, it sends data directly to their platform. The data they send, when it is not a victim of signal loss, is completely outside of the digital property owner’s control. This is precisely why stringent default browser security settings are on the rise.