Smart Bidding is Google's name for the set of automated bidding strategies that use machine learning to decide, auction by auction, how much to offer for each click. Instead of a person manually setting how much they're willing to pay per keyword, the system adjusts the bid in real time based on dozens of signals: device, time of day, location, user history, and many more that aren't even visible to whoever manages the account.
The most common strategies and what each one is for
Maximise conversions, which aims to get as many conversions as possible with the available budget, without setting a target cost. Target CPA (cost per acquisition), which tries to keep the average cost per conversion close to a specific figure you define. Target ROAS (return on ad spend), designed mainly for ecommerce, which prioritises maximising sales value relative to spend, not just the number of conversions. Picking the wrong strategy for the business's real goal is one of the most expensive and most frequent mistakes in Google Ads accounts.
Why it needs data to work well
These algorithms learn from past results, so a brand-new campaign with no conversion history has nothing to train the system on, and the first few days or weeks (what Google calls the learning phase) tend to show unstable results while the algorithm calibrates. Constantly switching bidding strategies, or changing the target every few days, resets part of that learning and tends to make performance worse rather than better.
When manual bidding still makes sense
In campaigns with very low conversion volume (where the algorithm doesn't have enough data to learn reliably), in launches where you need tight control over spend while validating whether the product or service works at all, or in accounts where the business goal is too specific to fit any of the standard metrics Smart Bidding offers. Outside those cases, manual bidding rarely beats automated bidding over the medium term today.
What the account manager can still do once they no longer pick the bid
Control shifts elsewhere: the quality of conversion tracking (if the system measures what counts as a real conversion badly, it will optimise towards the wrong goal), the quality of the creatives and audience targeting, and correctly setting the CPA or ROAS target based on the business's real margin. Managing an account well with Smart Bidding isn't "letting the machine do it alone", it's spending your time on the variables you can actually control.
Frequently asked questions
Does Smart Bidding work equally well for any budget?
It needs a minimum monthly conversion volume to learn reliably (Google usually recommends at least 30 conversions a month per strategy). With very small budgets, it can take longer to stabilise, or never fully do so.
Can I combine manual and automated bidding in the same account?
Yes, it's common to use manual bidding on low-volume or test campaigns, and automated bidding on mature campaigns with enough history, within the same ad account.
What happens if I change the CPA or ROAS target frequently?
Every significant change can re-trigger part of the algorithm's learning phase, causing temporary instability in results. It's better to make target changes sparingly and wait at least one or two weeks before evaluating the effect.