Why Google's AI Can Outbid You at Your Own Auction
Every time a Google ad could appear, an auction fires behind the scenes in milliseconds. Dozens of signals run simultaneously — device type, time of day, location, past search behavior, competitor bids. No human can process all of that and adjust a bid in real time. Smart Bidding is Google's machine learning system that does exactly that, setting a unique bid for every single auction. According to Google's own published data, the number of signals considered can reach into the hundreds of millions.
The Four Core Smart Bidding Strategies
- Maximize Conversions: Spends your full budget to get as many conversions (purchases, form fills, calls) as possible. You don't set a target number — Google optimizes within your budget. Good for new accounts or campaigns where you're still learning.
- Target CPA (Cost Per Acquisition): You tell Google the maximum you're willing to pay for each conversion or lead. Google tries to stay as close to that number as possible. Use this once you know what a customer is worth and your budget can support it.
- Target ROAS (Return on Ad Spend): You define how much revenue you want for every lira spent. Makes the most sense for e-commerce and businesses selling products at different price points. Requires conversion value tracking through GA4 to be active.
- Maximize Clicks: Technically outside the core Smart Bidding family — it's rule-based, not AI-driven. It delivers the most clicks but isn't optimized for conversions. Mostly useful for brand awareness or pure traffic goals.
At least 30–50 conversions
Google's recommended monthly minimum for Smart Bidding to work reliably. Below this threshold the algorithm lacks enough data and predictions become unreliable. (Source: Google Ads Help Centre official guidance)
Switching to Target CPA or Target ROAS before you have enough conversion data usually backfires. Start with Maximize Conversions to build up the signal base, then graduate to goal-based strategies. Rushing the transition often wastes budget without hitting the target.
After switching strategies, give the campaign at least 2–4 weeks. During this learning period Google's algorithm adjusts to the new goal. Performance may dip — reverting the change immediately resets the learning and wastes the data already collected.