Smart Bidding strategies have been available long enough that most advertisers have formed strong opinions about them. Some swear by them. Others have sworn off them after a bad experience. Very few have a clear-eyed understanding of what they actually do and what conditions they require to work.
Myth 1: Smart Bidding will optimise its way out of any situation
The most common misapplication is treating Smart Bidding as a solution to a structural problem. An account with poor conversion tracking, insufficient conversion volume, or misaligned campaign structure won't be rescued by Target CPA or Target ROAS.
Target CPA needs a minimum of 30-50 conversions per campaign per month to function reliably. Below that threshold, the algorithm guesses - and its guesses aren't better than manual bidding at low data volumes.
Myth 2: Maximise Conversions means maximum conversions at any cost
It does - and that's exactly the problem when used without a tCPA constraint in a campaign with a real budget limit. Maximise Conversions will spend your entire daily budget to get as many conversions as possible. If the resulting CPAs don't stack up commercially, the strategy is working as designed. It's just not working for your business.
Myth 3: Switch strategies when performance dips
Smart Bidding strategies have learning periods - typically 2-4 weeks after any change. During that window, performance is often worse than it will be once the algorithm has settled. Switching strategies when a dip appears resets the learning period and means you never find out what the strategy would have achieved.
The practical rule: give any Smart Bidding change at least 4-6 weeks of stable performance data before evaluating whether it's working.
What actually works
Smart Bidding works best in accounts with high conversion volumes, clean conversion tracking, consolidated campaign structures, and realistic targets. The work is to build the conditions first. The bidding strategy follows.