How do you improve performance through Google machine learning?

Some Google Ads optimisations are tedious but crucial to the success of a campaign, such as the manual optimisation of bids. How about automating this type of task more efficiently, allowing you to focus on more strategic actions?
Google machine learning offers you these benefits: increasing productivity, boosting your growth and reaching your goals quickly.

Posted by Laure Salber
on 25/10/2018

Google’s automation: new optimisations and growth opportunities

Automation on Google Ads is based on 3 features:

  • Contextual signals: every day Google aggregates billions data on its users behaviour and on contextual signals, such as a user’s geographic location, audiences, device, browser and time of day.
  • Smart Bidding strategies use these contextual signals to identify conversion opportunities. The system analyses them and searches for the most relevant combinations to automatically set the most appropriate bid for each auction in real time. As a human being, it would be impossible to combine so much data and decide on the value of each auction in just a second!
  • Learning from queries: smart bidding maximises the performance of these new keywords generating a low volume of traffic. To do so, this system uses similar auctions in the account to set the best possible bid, while reducing fluctuations in performance.

How do you set it up?

Smart bidding is accessible to every advertiser, irrespective of spend. The only requirement is a minimum number of conversions in the last 30 days to provide sufficient data to the machine learning algorithms to set bids efficiently. It is recommended to have at least 15 conversions to use the “maximise conversions” strategy, 30 conversions for Target CPA and 50 conversions for Target ROAS.

A few tips:

  • Do not let campaigns get restricted by budget. Let the system take advantage of the full potential to obtain a better ROI.
  • Simplify the structure of campaigns. Even though Google’s automation is efficient on low volume, it performs much better on campaigns with a consolidated rather than granular structure.
  • Test wider keyword targeting: you provide more data to the tool. However, make sure you check the search terms to ensure the relevance of the queries. Don’t completely give up on other types of targeting that allow you to reach a more precise audience.
  • Add audience lists at campaign or ad group level without applying any auction adjustments. Smart bidding will optimize bids based on the performance of the lists.
  • Choose the attribution model best suited to your type of conversion. By default, Google Ads applies the “last-click” attribution model. You may want to change this in order to optimise campaigns more efficiently.
  • Switch to manual bidding during strategic times, such as sales, Christmas or Valentine’s Day. Smart bidding strategies are not necessarily adequate in these cases.
  • Entrust the management of your campaigns to a Google Partner Premier agency. A good command of the subject is essential for the success of your campaigns. Certified administrators will be able to guide you to the strategy best suited to your targets.


Smart bidding strategies have a learning phase of about 5 weeks. It is recommended to avoid making major changes during this period, as the resulting analysis may be flawed.

Posted in : Google AdWords

Let's get in touch


    Subscribe to our newsletter (in french only).