Using Adobe Analytics as the behavioral data source lets clients use the view-based and/or purchase-based behavioral data from Analytics in Adobe Target Recommendations activities. This feature is especially helpful in situations where the Target Recommendations setup is new and Analytics has much historical data to use.
Using Analytics as the behavioral data source can act as a rich source of information about user behavior. This information might include data from a third-party source or feed that is shared only with Analytics.
While creating criteria in Recommendations, there are two radio buttons that let you choose which data source is to be used: mboxes or Analytics. To create a criteria, click Recommendations > Criteria > Create Criteria > Create Criteria. For more information, see Create criteria.
If these two buttons do not display in your account, reach out to Customer Care.
Using Analytics as the behavioral data source for recommendations also lets you deploy specific use cases without the requirement of tagging entity pages with all the Target entity parameters. Although that requires certain pre-requisites to be in place, availability of “Product Variables” is the most important thing for that functionality to work seamlessly. Regular eVars and Props are not sufficient for this handshake to happen automatically between Analytics and Target.
You can use Analytics as the behavioral data source to:
The following sections help you implement this feature on the Analytics side.
Implement product variables in Analytics with the necessary attributes that are required for Target Recommendations.
A Target Recommendations sample feed format acts as guide on which all attributes must be defined in the product variables. Later those values must be “mapped” within the Target UI for the respective Target entity values.
If it is a content site, the respective content pieces must be treated as “products” and associated attributes about that content must be passed as attributes. Such attributes can include author name, publish date, content title, month of release, and so forth. Granularity of category level, or category types, should be decided by the business based on use-case requirements.
For more details on how to set up product variables, see products in the Implement Adobe Analytics guide. Some of the notes in that documentation need discretion of the team who is deploying it (example : Category). It is always advised to consult with Adobe before doing this activity.
Analytics data is sent via a daily feed. Behavioral results can take up to 24 hours to be reflected within recommendations results on your site. As with all Recommendations criteria settings, this data source can and should be tested.
For quick decision making on which data source is to be used, if there is much organic data generated every day by users, and not much dependency required on historic data, then using a Target mbox as the behavioral data source can be a good fit. In cases of less availability of organic data generated recently, if you want to bank upon Analytics data, then the using Analytics as the behavioral data source is a good fit.
Now it is time to map these variables on Target side for continuous supply of behavioral data.
In Target, click Recommendations, then click the Feeds tab.
Click Create Feed.
Select Analytics Classifications, then specify the report suite.
Click Next to advance to the Schedule settings, the select a frequency period for the feed:
You can also select the time of day for the feed to process.
Click Next to advance to the Mapping settings, then map the field column headers to the appropriate Recommendations field names.
Consider the following FAQs as you use Analytics with Target:
entity.categoryIdvalues required to be passed within the Target mbox call?
Yes, those two values are still required. The rest of the attributes can be passed via an Analytics feed, as discussed in this document.
Yes, you can. The method is similar when using Target stand-alone. In this case, however, you must be mindful about the timing factor. The entity variables that are supposed to match with the profile variables depend on the data layer that might appear much later on the page.