Key Takeaways

1. The Power of Smart Bidding:

   • Dramatic improvements possible: In one case, 97% decrease in cost per conversion, 3,282% increase in conversion rate, and 15,100% increase in conversions in 32 days.

2. Core Principles of Smart Bidding Success:

   • Feed the beast: Provide ample, quality data to the algorithm.
   • Exclude permanently: Use data exclusions wisely and don't remove them.
   • Model extensively: Understand and work with the underlying models.

3. Bayes' Theorem:

   • Smart Bidding uses Bayesian probability, continuously updating its beliefs based on new data.

4. Pinning Ads:

   • Pin ads only when necessary, as it can limit the algorithm's learning and optimization.

5. ARIMA Models:

   • Google uses ARIMA (Autoregressive Integrated Moving Average) models for forecasting.

6. Common Pitfalls:

   • Google Ads can't see "All Conversions" - ensure it has visibility into the right data.
   • Messing with historical data can confuse the algorithm.
   • Not allowing Google enough time to learn can lead to suboptimal performance.

7. Modeling the Model:

   • Understanding and modeling Google's approach can lead to better strategy and results.

Action Items

1. Review Data Feeding:

   • Ensure you're providing comprehensive, quality data to the Smart Bidding algorithm.
   • Check that all relevant conversion actions are visible to Google Ads.

2. Audit Data Exclusions:

   • Review your current data exclusions.
   • Avoid removing existing exclusions as it can disrupt the algorithm's learning.

3. Experiment with Unpinned Ads:

   • Test campaigns with unpinned ads to allow for more algorithmic learning.

4. Explore ARIMA Models:

   • Search for "ARIMA_PLUS" in Google Ads to understand how Google is forecasting for your campaigns.

5. Allow Learning Time:

   • Be patient and give Google's algorithm time to learn and optimize before making major changes.

6. Avoid Historical Data Manipulation:

   • Refrain from altering past data, as it can confuse the Smart Bidding algorithm.

7. Model Your Approach:

   • Develop your own models to understand and predict Smart Bidding behavior.
   • Use this understanding to inform your bidding strategies and optimizations.

8. Continuous Learning:

   • Stay updated on Smart Bidding developments and best practices.
   • Consider downloading the presenter's Data Exclusions slides for more in-depth information.

Remember: Smart Bidding is powerful but requires a deep understanding of its underlying principles and careful management of data inputs. Success comes from working with the algorithm, not against it.

For more information:

• LinkedIn: Nathan Ifill
• Twitter: @nathanifill
• Presentation Deck: speakerdeck.com/nathanifill