Machine learning optimization leverages machine learning and a unique algorithm to maximize a campaign's performance. Machine learning operates in two distinct modes:
Learning mode: In learning mode, the technology captures information to create a model. The algorithm requires 150 clicks or video views to create an initial model. In this stage, the campaign uses its Default Bid amount to bid on all impressions.
Optimized mode: In optimized mode, the technology uses an available model to decide if the DSP should bid on an impression or not, as well as how much to bid. The campaign uses the KPI Goal value and the Maximum Bid to determine the bid price based on the overall historical probability of achieving the event that affects the KPI (e.g. a click, a view). Bid prices are unique for every impression and depend on many factors, including the inventory available, the timing of the impression, and the device that will display the impression.
The campaign's optimization status (LEARN or OPT) appears in the campaign tab in the Groups view.
To make sure that there are enough events for the model to work as desired, we recommend a minimum spend of $5,000 per brand over the course of 2 weeks.
Note: The optimization technology constantly refines its model to try to achieve your desired KPI. However, you should still check your campaign regularly to be sure that it achieves the results you are looking for.
Enabling Machine Learning Optimization for a Campaign
Create or edit a campaign.
Make sure that the campaign has a Start date and an End date. If the group specifies flight dates, campaign flight dates are optional.
In the Optimization section, select Machine learning optimization.
- Select a Goal from the dropdown menu and enter a goal value:
- Cost Per Click (eCPC): Optimize for clicks on image ads. Enter the target cost per click that you want your campaign to achieve.
- Cost Per Acquisition (eCPA): Optimize for cost per conversions. Enter the target cost that you want to spend per acquisition for this campaign.
- Click-Thru Rate (CTR): Optimize for number of clicks per number of served impressions. Enter the target percentage of clicks that you want the campaign to achieve (between 0.01% and 10.00%).
- Cost Per Completed Video (eCPCV): Optimize for views of video ads. Enter the target cost per completed video view that you want your campaign to achieve.
- Viewable Rate (VR): Optimize for viewable impressions. Enter the minimum percentage of viewable impressions that you want your campaign to bid on. The value must be a multiple of 10 between 10 and 90.
- Viewable CPM (VCPM): Optimize for the cost of viewable impressions. Enter the target bid that you want to spend per 1000 viewable impressions.
- Video Completion Rate (VCR): Optimize for complete video views, expressed as a percentage: (Number of Completed Videos / Number of Video Starts) * 100. Enter the percentage of video completes that you want to achieve.
Note: If you select Viewable Rate or Viewable CPM, Measure Viewability automatically turns on for this campaign, and you cannot disable it while you have these goals selected. For more information about these goals and how we measure viewability, see Viewability.
Enter the Maximum Bid, which is the highest CPM bid that can be given to a placement.
Note: The maximum CPM bid does not have to be equal or greater than the campaign's Default CPM Bid. We recommend starting with a Maximum Bid equal to the campaign's Default CPM Bid. After the campaign has run in optimized mode for a few days, you can lower the Maximum Bid if it's not achieving the desired CPC goal, or you can leave it unchanged if it is achieving the goal.
In the Inventory tab, choose the exchanges that you want the campaign to run on. You can set optimized campaigns to Run on Selected Exchanges or Run on All Exchanges.
Other Optimization Types
You can't use algorithmic optimization and machine learning optimization at the same time.
You can use ad optimization while using machine learning optimization.