We Analyzed 305 Google Ads Campaigns... Here’s What We Found

Google Ads made a significant change on August 17th, 2026, altering how smart bidding functions, particularly for target CPA and target ROAS when budgets were limited. While Google promised this would enable scaling with confidence, it came with a crucial caveat: sufficient data within individual campaigns. Overly complex account structures often limit this essential conversion data, leading to unpredictable results and hindering the promised scalability. The speaker shares insights from analyzing 305 campaigns before and after this date, focusing on data volatility. The core finding is that more conversion data within individual campaigns leads to more predictable outcomes. This is crucial for businesses spending significant amounts on Google Ads, as it impacts their ability to confidently scale budgets. The August 17th change, a platform-wide event affecting all advertisers, removed a loophole where limited-budget campaigns with a target ROAS of, say, 200% might perform at 400-500%. Now, the algorithm strictly targets the set ROAS, causing performance drops for those who hadn't set realistic targets. The speaker emphasizes that the focus shouldn't be on the immediate performance drop but on ensuring campaigns have the necessary data to adapt to these bidding changes. Google Ads, being primarily a demand-capture platform based on keyword intent, offers a degree of predictability that other platforms, like Meta, which can be supercharged by creative, may not offer in the same way. Both platforms are important, and a balanced approach is key. The analysis of 305 e-commerce campaigns across Australia, the US, and Europe revealed that campaigns with fewer than seven conversions a week experienced significant volatility, with 50% seeing over a 50% change in ROI, mostly negative. As conversion volume increased, predictability improved: 25% volatility for 7-30 conversions per week, and 13% for 30-50 per week. The "golden number" identified is 50-plus conversions per week per campaign. Across all 305 campaigns, zero experienced more than a 50% variance when they achieved this benchmark. This predictability is linked to Google Ads' smart bidding recommendations: at least 30 conversions a month for basic smart bidding, and 50 conversions a month before using target ROAS or CPA. Furthermore, a campaign exits its learning phase upon reaching 50 conversions or after three conversion cycles. Achieving at least 50 conversions weekly allows campaigns to exit learning faster, enabling more aggressive budget increases and quicker adaptation to changes. The speaker advocates for a "less is more" approach to account structure, avoiding over-optimization and overly complicated setups. By leveraging conversion data and aiming for condensed structures with a minimum of 50 conversions per campaign weekly, advertisers can achieve better manageability and predictable scaling. The ideal starting point for account structure is one campaign and one ad group, forcing justification for any additions. This strategy ensures each component has a clear goal, leading to more assured outcomes.