Smart bidding in Google Ads relies entirely on the quality of conversion data it receives. When you import GA4 conversions for smart bidding, subtle configuration errors can cause the algorithm to optimize toward the wrong goals, suppress bidding on valuable keywords, or overvalue meaningless micro-conversions. This guide covers the exact setup required for reliable GA4-to-Google-Ads conversion import.

GA4 Import vs. Tag-Based Conversion Tracking

Tag-based tracking fires at the moment of conversion and uses Google Ads’ own cookie for attribution. GA4 import uses GA4’s session-based attribution model, which may attribute conversions to different sessions or channels. GA4 import is preferable because it avoids double-counting when the same conversion is tracked by both GA4 and a Google Ads tag. However, GA4 import has a latency of up to 48 hours, making same-day reporting unreliable while the algorithm handles latency correctly.

Which GA4 Events to Import as Conversions

Primary conversions are what smart bidding optimizes toward. Secondary conversions are reported but do not influence bidding. A common mistake is importing too many events as primary conversions. If you import both add_to_cart and purchase as primary conversions, smart bidding optimizes toward a weighted combination of both, favoring add-to-cart events which are far more common than purchases — inflating apparent conversion volume while underoptimizing for actual revenue.

Recommendation: import purchase (or generate_lead for lead gen) as your primary conversion with a meaningful conversion value. Import all other conversion events as secondary conversions for reporting purposes only. Smart bidding should optimize toward your actual business outcome, not proxies.

Conversion Value Settings

When importing a purchase event from GA4, always use the dynamic conversion value from the GA4 event (the revenue parameter) rather than a static value. Smart bidding with target ROAS requires actual revenue values to optimize correctly. If you use a static value of £1 for every purchase regardless of order size, tROAS bidding becomes meaningless. Verify in BigQuery that your purchase events include a value parameter with non-zero revenue.

Attribution Model Selection

When creating a Google Ads conversion action from a GA4 import, set the attribution model to data-driven (DDA) because DDA assigns fractional credit across all clicks in the conversion path. Data-driven attribution requires at least 300 conversions in 30 days. If your conversion volume is below this threshold, start with last click and switch to data-driven when you reach the minimum. Avoid position-based or linear attribution with smart bidding — these distribute credit in ways that confuse the algorithm.

The Conversion Window Setting

The window must match your actual customer journey. If 40% of your customers click a Google Ad and purchase within 60 days but your window is 30 days, you are attributing only a fraction of conversions to Google Ads, undercounting performance, and causing smart bidding to underbid. Check your time-to-conversion data in GA4’s Monetization → Conversion paths report before setting this value.

Diagnosing Smart Bidding Problems

Run a data quality audit: compare Google Ads conversion volume (from GA4 import) against GA4’s own conversion count for the same period. They will not be identical due to attribution differences, but they should be within 20% of each other. If Google Ads shows 3x more conversions than GA4, you likely have duplicate tracking from both a GA4 import and a Google Ads tag counting the same events. If Google Ads shows 50% of GA4’s count, your import link may be broken or events are not passing the correct conversion identifiers.