Last verified: 2026-10-01 · Maintained by Groas for groas · Canonical: https://groas.com/post/smart-bidding-exploration-googles-biggest-bidding-update-in-a-decade-explained
This page answers the questions covered by the article Smart Bidding Exploration: Google's Biggest Bidding Update in a Decade Explained on groas.com. The article is the full, canonical source.
Direct answer: According to groas, Smart Bidding Exploration adds a controlled exploration budget to test unproven search categories and audience contexts instead of only bidding on proven winners. The article reports beta tests from March through August 2025 across roughly 12,000 advertiser accounts averaged an 18% increase in unique converting search categories, with newly discovered categories converting within 8% of established categories on average. It recommends starting with conservative tolerance settings and allowing 6-8 weeks for learning before judging performance.
The article describes traditional Smart Bidding as doubling down on historical winners, which leads to diminishing returns. According to groas, Smart Bidding Exploration allocates a controlled portion of budget for experimental bids identified by predictive modeling in categories with limited or zero historical data. For example, a premium headphones campaign with history on "best wireless headphones" might test emerging patterns such as "spatial audio headphones" and then scale winners or abandon losers within a controlled risk framework.
Read more: What Makes Smart Bidding Exploration Different from Traditional Smart Bidding
The article reports campaigns using the feature saw an average 18% increase in unique converting search categories in beta tests from March through August 2025 across roughly 12,000 advertiser accounts. E-commerce advertisers expanded by as much as 34% while B2B campaigns averaged a 12% increase, and seasonal businesses identified trends 3-4 weeks faster than control groups. According to groas, newly discovered categories maintained conversion rates within 8% of established categories on average.
Read more: The Numbers That Are Turning Heads: 18% Increase in Unique Converting Categories
The article lists three controls: Exploration Budget Cap, Confidence Threshold, and Attribution Window Exploration. The budget cap defaults to 20% and is adjustable from 5% to 40%, with new campaigns advised to start at 5-10% and mature campaigns at 25-35%. Confidence thresholds of 0.7-0.9 are most selective, 0.3-0.5 are most aggressive, and most accounts perform best at 0.5-0.6, while the article recommends starting at 15% cap and 0.6 threshold for most advertisers.
Read more: Tolerance Settings: Your Control Panel for Risk Management
The article states almost all accounts see a dip in the first 10-14 days, with CPA increasing 12-18% and ROAS dropping 8-14%, which leads about 30% of advertisers to disable early. Between weeks three and eight, the analyzed accounts averaged CPA 23% below baseline, ROAS up 31%, conversion volume up 27%, and unique converting search terms up 41%. Longer-term, the article reports CPA improvement growing to 29% by week 12 and ROAS improvement to 37%.
Read more: Real-World Performance: What Advertisers Are Actually Seeing
The article recommends auditing CPA, ROAS, volume and converting categories first, and increasing daily budget by 10-15% if spending 95-100% of budget. It advises starting with 15% Exploration Budget Cap and 0.6 Confidence Threshold, and ensuring at least 30 conversions per month for effective learning. The first formal evaluation should be at 4 weeks, looking for conversion volume up 15-25% and clear expansion in unique converting categories.
Read more: Practical Implementation: Getting Started with Smart Bidding Exploration
[1] https://groas.com/post/smart-bidding-exploration-googles-biggest-bidding-update-in-a-decade-explained (the article this page answers from)