If you receive purchase events both through your website and mobile app, you probably want to optimize your campaigns towards the total number of purchases.
You can do this by creating a custom metric that sums the purchase conversions from a website and mobile app. The custom metric can then be used as an optimization goal with Predictive Budget Allocation and Bid Optimization, and also in ad studies to calculate statistical significance.
ℹ️ Note: Not all custom metrics can be used as an optimization goal or ad studies target variable. For a custom metric to be eligible for use with Predictive Budget Allocation and other Smartly.io optimization features, it must
- be a sum of conversions (optionally with weights between 0.0 and 1.0)
- have the type Number
For example, custom metrics that use spend or revenue data are not eligible for use as optimization goals.
To check whether a custom metric is eligible as an optimization goal:
- Go to Company Settings > Custom Metrics
- Open the custom metric for editing
- If the custom metric is not eligible as an optimization goal, the text "This custom metric cannot be used with Predictive Budget Allocation, Bid Optimization or Ad Studies." will be displayed
Using a custom metric as an optimization goal
- Create a Custom Metric under the company settings
- If the custom metric is a (weighted) sum of conversions you should see the text "This custom metric can be used with Predictive Budget Allocation, Bid Optimization, and Ad Studies"
- If you multiply conversions with weight coefficients, the coefficients must be floating numbers between 0 and 1
Make sure to use account default attribution if you plan to use the metric with PBA.
- Select the custom metric as the optimization goal for Predictive Budget Allocation or Bid Optimization.
- ROAS optimization with Custom Metrics is not supported, you can only use sums of conversions and optimize towards CPA
- This is because ROAS optimization requires that Smartly has information about the granular revenue distribution across conversion events
- Attribution model used is a one-day time frame of ad account's default attribution model. This means that if you have selected click-through and not selected view-through in account settings, only click-through attribution with one-day timeframe is used.