Adobe Experience Platform enables you to bring data fragments together from multiple sources and combine them in order to see a complete view of each of your individual customers. When bringing this data together, merge policies are the rules that Platform uses to determine how data will be prioritized and what data will be combined to create a unified view.
For example, if a customer interacts with your brand across several channels, your organization will have multiple profile fragments related to that single customer appearing in multiple datasets. When these fragments are ingested into Platform, they are merged together in order to create a single profile for that customer. When the data from multiple sources conflicts (for example one fragment lists the customer as “single” while the other lists the customer as “married”) the merge policy determines which information to include in the profile for the individual.
Using RESTful APIs or the user interface, you can create new merge policies, manage existing policies, and set a default merge policy for your organization. This guide provides step-by-step instructions for working with merge policies using the Adobe Experience Platform user interface (UI).
If you would prefer to work with merge policies using the Real-time Customer Profile API, please follow the instructions outlined in the merge policies API guide.
This guide requires a working understanding of several important Experience Platform features. Before following this guide or using Profile APIs, please review the documentation for the following services:
Each profile fragment contains information for just one identity out of the total number of identities that could exist for an individual. When merging that data together to form a customer profile, there is the potential for that information to conflict and priority must be specified. Selecting a merge method allows you to specify which dataset attributes to prioritize if a merge conflict occurs between datasets.
There are two possible merge methods available for merge policies. Each of these methods are summarized below with additional details provided in the sections that follow:
As profile records are ingested into Experience Platform, a system timestamp is obtained at the time of ingestion and added to the record. When Timestamp ordered is selected as the merge method for a merge policy, profiles are merged based on the system timestamp. In other words, merging is done based on the timestamp for when the record was ingested into Platform.
Occasionally there may be use cases where it is necessary to supply a custom timestamp and have the merge policy honor the custom timestamp rather than the system timestamp. Examples of this include backfilling data or ensuring the correct order of events if records are ingested out of order.
In order to use a custom timestamp, the External Source System Audit Details schema field group must be added to your Profile schema. Once added, the custom timestamp can be populated using the
lastUpdatedDate field. When a record is ingested with the
lastUpdatedDate field populated, Experience Platform will use that field to merge records across datasets. If
lastUpdatedDate is not present, or not populated, Platform will continue to use the system timestamp.
You must ensure that the
lastUpdatedDate timestamp is populated when ingesting an update on the same record.
The following screenshot displays the fields in the External Source System Audit Details field group. For step-by-step instructions on working with schemas using the Platform UI, including how to add field groups to schemas, please visit the tutorial for creating a schema using the UI.
To work with custom timestamps using the API, refer to the merge policies endpoint guide section on using custom timestamps.
When Dataset precedence is selected as the merge method for a merge policy, you are able to give priority to profile fragments based on the dataset from which they came. An example use case would be if your organization had information present in one dataset that is preferred or trusted over data in another dataset.
In order to create a merge policy using Dataset precedence, you must select the Profile and ExperienceEvent datasets that are included and then you can manually order the Profile datasets for precedence. Once the datasets have been selected and ordered, the top dataset will be given highest priority, the second dataset will be second-highest, and so on.
Identity stitching (ID stitching) is the process of identifying data fragments and combining them together to form a complete profile record. To help illustrate the different stitching behaviors, consider a single customer who interacts with a brand using two different email addresses.
To learn more about identities and their role in generating profiles and segments, please begin by reading the Identity Service overview.
An organization can create a default merge policy for their organization to use when merging profile fragments. This allows users to easily select the default policy when performing actions in Experience Platform such as viewing customer profiles or creating segments. In most cases, unless another merge policy is specified, the default merge policy will be used.
Each organization can create multiple merge policies related to a single XDM schema class, however they can only have one default merge policy declared for each class. For example, your organization could have a default merge policy related to the XDM Individual Profile class and a different default merge policy for a custom-built Product Inventory class.
If you create a new merge policy and set it as the default, the previous default merge policy will be automatically updated by the system to no longer be the default.
Profile counts and segments with an existing associated default merge policy may be affected. Any segment that has a default merge policy applied will be updated to the new default merge policy.
Within the Experience Platform UI, you can begin working with merge policies by selecting Profiles in the left navigation and then selecting the Merge Policies tab. This tab includes a list of all existing merge policies for your organization, as well as details for each merge policy including the policy name, whether or not the merge policy is the default merge policy, and the schema class to which the merge policy relates.
To select which details are visible, or to add additional columns to the display, select Configure columns and click on a column name to add or remove it from view.
To create a new merge policy, select Create merge policy on the merge policies tab.
On the New merge policy workflow screen, you can provide important information for your new merge policy through a series of guided steps.
The first step in the workflow allows you to configure your merge policy by providing basic information. This information includes:
Once the required fields have been completed, you can select Next to continue with the workflow.
When creating or editing a merge policy, you can view the union schema for the chosen schema class by selecting View Union Schema.
This opens the View Union Schema dialog, showing all contributing schemas, identities, and relationships associated with the union schema. You can use the dialog to explore the union schema in the same way that you would by accessing the Union Schema tab in the Profiles section of the Platform UI.
For detailed information on union schemas, including how to interact with them in the Union Schema tab or the View Union Schema dialog shown in the merge policies workflow, please visit the union schema UI guide.
On the Select Profile datasets screen, you must select the Merge method that you wish to use for your merge policy. Also displayed on the screen is the total number of Profile datasets in your organization that relate to the schema class that was selected on the previous screen.
Depending on the merge method that you choose, all of the Profile datasets will be merged by the order in which they were last updated (timestamp ordered) or you will need to select which Profile datasets to include in the merge policy and the order in which to merge them (dataset precedence). For more information on merge methods, please review the merge methods section provided earlier in this document.
Selecting Timestamp ordered as the merge method means that attributes from the most recently updated datasets will take precedence. This applies across all Profile datasets.
Selecting Dataset precedence as the merge method requires you to select Profile datasets and manually prioritize them. Each dataset listed also includes the status of the last batch ingested or displays a notice that no batches has been ingested into that dataset.
You can select up to 50 datasets from the dataset list to include in the merge policy. As datasets are selected, they are added to the Select datasets section, allowing you to drag and drop the datasets and order them according to your desired precedence. As the datasets are adjusted in the list, the ordinal (1, 2, 3, etc) next to the dataset will update, displaying priority (1 being given the highest priority, then 2, and onward).
Selecting a dataset also updates the Union schema section, showing the fields in the union schema to which each dataset contributes data. For more information on union schemas, including how to interact with the visualizations in the UI, please reference the union schema UI guide
The next step in the workflow requires you to select ExperienceEvent datasets. This screen is influenced by the merge method that you selected on the Select Profile datasets screen.
Also displayed on this screen is the total number of ExperienceEvent datasets created by your organization related to the schema class that you selected on the merge policy configuration screen.
If you selected Timestamp ordered as the merge method for Profile datasets, the attributes from the most recently updated ExperienceEvent datasets will take precedence here as well.
If you selected Dataset precedence as the merge method for Profile datasets, you will need to select ExperienceEvent datasets to include. You can select up to 50 ExperienceEvent datasets from the dataset list. As datasets are selected, they appear in the Select datasets section.
ExperienceEvent datasets cannot be manually ordered, instead the attributes in the ExperienceEvent datasets are appended to the Profile datasets if they are part of the same profile fragment.
Similar to selecting Profile datasets, selecting an ExperienceEvent dataset also updates the Union schema section, showing the fields in the union schema to which each dataset contributes data. For more information on union schemas, including how to interact with the visualizations in the UI, please reference the union schema UI guide
The final step in the workflow is to review your merge policy. The Review screen displays the name of your new merge policy, the schema class on which it is based, the ID stitching option that you selected, as well as the merge method and the datasets included in the merge policy. To view all of the Profile or ExperienceEvent datasets included, select the number of datasets to expand the dropdown list.
Please ensure that you review your merge policy carefully before selecting Finish to complete the creation workflow.
If you selected Timestamp ordered as the merge method for your merge policy, the list of Profile datasets includes all of the datasets that have been created by your organization related to the schema class, in order of timestamp. The list of ExperienceEvent datasets includes all datasets that your organization has created for the chosen schema class and will be appended to the Profile datasets.
If you selected Dataset precedence as the merge method for your merge policy, the lists of Profile and ExperienceEvent datasets include only the Profile and ExperienceEvent datasets that you selected during the creation workflow, respectively. The order of the Profile datasets should match the precedence that you specified during creation. If it doesn’t, use the Back button to return to the previous workflow steps and adjust the priority.
After completing the workflow to create a new merge policy, you are returned to the Merge Policies tab. The list of merge policies for your organization should now include the merge policy that you just created.
From the Merge Policies tab, you can modify an existing merge policy created for the XDM Individual Profile class by selecting the Policy name for the merge policy you wish to edit.
When the Edit merge policy screen appears, you can make changes to the name and ID stitching, as well as change whether or not this policy is the default merge policy for your organization.
Select Next to continue through the merge policy workflow to update the merge method and datasets included in the merge policy.
Once you have made the necessary changes, review your merge policy and select Finish to return to the Merge policies tab.
Changing a merge policy can affect segmentation and profile results, as it will alter the way in which data conflicts are resolved.
When creating or updating a merge policy, a check is performed to determine if the merge policy violates any of the data usage policies defined by your organization. Data usage policies are part of Adobe Experience Platform Data Governance and are rules that describe the kinds of marketing actions that you are allowed to, or restricted from, performing on specific Platform data. For example, if a merge policy was used to create a segment that activated to a third-party destination, and your organization had a data usage policy preventing the export of specific data to third parties, you would receive a Data governance policy violation detected notification when attempting to save your merge policy.
This notification includes a list of data usage policies that have been violated and allows you to view the details of the violation by selecting a policy from the list. Upon selecting a violated policy, the Data lineage tab provides the reason for the violation and the affected activations, each providing more detail into how the data usage policy has been violated.
To learn more about how data governance is performed within Adobe Experience Platform, please begin by reading the Data Governance overview.
Now that you have created and configured merge policies for your organization, you can use them to adjust the view of customer profiles within Platform and to create audience segments from your Profile data. See the Segmentation overview for more information on how to create and work with segments using the Experience Platform UI and APIs.