Enabling AI-driven advertising with paid click capture on Facebook Ads, Google Ads, and other platforms
Pega Customer Decision Hub campaigns that target paid audiences use Next-Best-Action to select the most relevant action for each individual. Paid Media Manager communicates that decision to ad platforms, and targets the individual with relevant ads. Regardless of whether the targeted individual responds positively (by clicking the ad) or negatively (by ignoring it), their behavior should be communicated back to Pega Customer Decision Hub, so that the AI models that govern targeting can learn from the interaction. To do this, you can configure both the paid destination and Paid Media Manager to capture paid clicks.
Sample use case
A marketing analyst configures an action promoting the Open Platinum 50K Bonus Credit Card to known individuals, including customers and prospects. Based on the next-best-action decision, Pega Customer Decision Hub creates a paid audience for the action, for example, a Facebook Custom Audience. The willingness to pay for an ad for a given action to be shown to an individual is designated by the audience and aligned with the priority for that action for that individual. The priority decision can be based on things such as eligibility, propensity, or predicted value for the action per individual. For more information about the mechanism behind paid audience updates, see Extending next-best-action to digital advertising platforms with Paid Media Manager.
An individual named Troy Murphy is eligible, has a high propensity, and high predicted value for the Platinum Credit Card, so he is selected for a 50K bonus for that credit card. Troy's response to an ad for this action can help inform AI models.
Individuals' responses to actions can determine the propensity for the action across channels and also the effectiveness of the ad to drive the individual to complete the action. However, the response needs to be tied to the specific decision that chose this action for the individual. This done by using an Interaction ID.
For that, Paid Media Manager leverages delayed learning. Delayed learning is a method of capturing response data for previous decisions, for example, the decision that results in Next-Best-Action selecting the Open Platinum 50K Bonus Credit Card for Troy. Delayed learning stores the latest Interaction ID for each action for each individual. This enables Paid Media Manager to retrieve the Interaction ID and close the loop when there is a response to an ad, for example, a click. If the individual does not respond within a week, the model records a negative response. This can be overridden with the paidmedia/ResponseWaitingTime dynamic system setting. For more information, see Delayed learning of adaptive models.
Obtaining the Interaction ID
To obtain the Interaction ID, Paid Media Manager must have the customer ID of the individual and the ID of the action for which the interaction was recorded. Depending on the destination type, this is handled in one of the following ways:
- For Facebook Ads and Google Ads, the click URL of an ad must contain the ad set or ad group ID. When the ID is returned after an individual clicks on an ad, Paid Media Manager checks the ad platform to determine which audience was used to target the ad.
- For other ad platforms, Pega Customer Decision Hub generates a unique identifier for each Next-Best-Action audience. This identifier is called the external audience key. When the ID is returned after an individual clicks on an ad, it tells Paid Media Manager which action the paid click is for.
Previous topic Generating Next-Best-Action web audiences with Paid Media Manager Next topic Enabling your website to capture paid clicks