What AI-driven bidding actually does
With manual bidding you set a maximum cost per click for a keyword, and that bid applies to every search that matches it. With Smart Bidding, Google sets the bid at auction time, for each individual search.
To do that it looks at signals attached to that search: the device, the location, the time of day, the language, the browser and operating system, whether the person is on one of your audience lists, and the words they typed. It predicts how likely that click is to convert, and how much the conversion is likely to be worth, then bids up or down.
No person can make that decision for every auction. That is the real advantage. The trade-off is that you control the goal and the inputs, not the bid.
The bid strategies and when to use each
The conversion-based strategies aim at one of two goals: the number of conversions, or their value. Google also offers automated strategies that do not use conversions at all.
- Maximize conversions: spends your daily budget to get as many conversions as possible. A good start for a new lead generation campaign with no history.
- Maximize conversions with a target CPA: the same, but aims at an average cost per conversion you set. Use it once the campaign has steady conversion data and you know what a lead is worth.
- Maximize conversion value: spends the budget to get the highest total value. Needs conversions with values, such as ecommerce revenue or scored leads.
- Maximize conversion value with a target ROAS: aims at a return on ad spend you set. Best for ecommerce with reliable revenue tracking.
- Maximize clicks and Target impression share: automated, but not conversion-based. Useful for traffic or visibility goals, not for leads or sales.
What it needs from you
Smart Bidding optimises towards whatever you count as a conversion. If that is wrong, it gets very good at the wrong thing.
- Count only real actions: a submitted form, a call over a set length, a booking or a purchase. Remove page views and button clicks from the primary conversion goal.
- Turn on enhanced conversions where your site and consent setup allow it, so conversions are still measured when cookies are limited.
- For lead generation, import qualified leads or sales from your CRM as offline conversions, so the system learns which clicks became customers, not just which filled in a form.
- Give conversions a value when they are not all equal. A value per lead type lets you move to value-based bidding later.
- Set a budget that can afford the target. A target cost per lead that the budget only buys a handful of times a month gives the system little to learn from.
Can it run without manual intervention?
It can set bids without you, but the account still needs managing. Smart Bidding does not choose your keywords, write your ads, exclude irrelevant searches, fix a broken tracking tag or decide what a customer is worth to you.
When the strategy is new or its settings change, Google shows a Learning status in the bid strategy column while it recalibrates. Making several large changes in a row keeps it learning, so change one thing at a time and give it time to settle.
Google provides two tools for unusual periods. Seasonality adjustments tell the system to expect a short change in conversion rate, such as a sale weekend. Data exclusions tell it to ignore a period when tracking broke, so bad data does not skew the bids.
A weekly routine that keeps it honest
The work moves from setting bids to checking inputs and results.
- Check that conversions are still recording, and that the numbers match your CRM or shop.
- Read the search terms report and add negative keywords for anything irrelevant.
- Compare cost per conversion or ROAS with the target, over a long enough window to cover your conversion lag.
- Look at lead quality, not just lead count. If cheaper leads are worse leads, feed that back through offline conversions.
- Move targets in small steps. A sharp cut to a CPA target can stop a campaign spending.
When manual or simpler bidding still makes sense
Very low conversion volume, broken or untrusted tracking, or a brand-new account with no data are all reasons to start with Maximize conversions or Maximize clicks and move to targets later.
If your business cannot measure conversions online at all, fix that first. Without conversion data there is nothing for the AI to learn from.