Computed attributes overview
Personalization based on user behavior is a key requirement for marketers to maximize the impact of personalization. For instance, personalizing marketing email with the most recently viewed product to drive conversion, or personalizing webpage based on total purchases made by users to drive retention.
Computed attributes help quickly convert profile behavioral data into aggregated values at the profile level without dependence on engineering resources for:
- Enabling targeted one-to-one or batch personalization with activation of behavioral aggregates to Real-Time Customer Data Platform destinations and usage in ۶Ƶ Journey Optimizer
- Simplified audience segmentation with storage of behavioral aggregates as profile attributes
- Standardization of aggregated profile behavioral data for usage across platforms and apps
- Better data management with consolidation of old profile events data into meaningful behavioral insights
These aggregates are computed based on Profile-enabled Experience Event datasets ingested into ۶Ƶ Experience Platform. Each computed attribute is a profile attribute created on your profile union schema, and is grouped under the “SystemComputedAttribute” field group in your union schema.
Sample use cases include:
- Personalizing marketing emails with total reward points to congratulate users on being promoted to a premium tier
- Personalizing communications to users based on purchase counts and frequency
- Personalizing retention emails based on subscription expiry dates
- Re-targeting users who viewed but did not purchase a product with the last viewed product
- Activating event aggregates through computed attributes to a downstream system using Real-Time CDP Destinations
- Collapsing multiple event-based audiences into a more condensed group of computed attributes
- Re-targeting unauthenticated users offsite using recent partner IDs from events
This guide will help you to better understand the role of computed attributes within Platform, in addition to explaining the basics of computed attributes.
Understanding computed attributes
۶Ƶ Experience Platform enables you to easily import and merge data from multiple sources in order to generate Real-Time Customer Profiles. Each profile contains important information related to an individual, such as their contact information, preferences, and purchase history, providing a 360-degree view of the customer.
Some of the information collected in the profile is easily understood when reading the data fields directly (for example, “first name”) whereas other data requires performing multiple calculations or relying on other fields and values in order to generate the information (for example, “lifetime purchase total”). To make this data easier to understand at a glance, Platform allows you to create computed attributes that automatically perform these references and calculations, returning the value in the appropriate field.
Computed attributes include creating an expression, or “rule”, that operates on incoming data and stores the resulting value in a profile attribute. Expressions can be defined in multiple different ways, allowing you to specify which events to aggregate on, aggregate functions, or the lookback durations.
Functions
Computed attributes let you define event aggregates in a self-serve manner by leveraging pre-defined functions. The details on these functions can be found below:
Minimum order amount in the last 4 weeks
Maximum order amount in the last 4 weeks
Lookback periods
Computed attributes are calculated in batches, letting you keep your aggregates fresh and using the latest events. In order to support these scenarios with minimal delay, the refresh frequency varies depending on the event lookback period.
The lookback period refers to the amount of time that is reviewed when aggregating Experience Events for the computed attribute. This period of time can defined in hours, days, weeks, or months.
The refresh frequency refers to the frequency that the computed attributes are refreshed. This value is dependent on the lookback period, and is automatically set.
For example, if your computed attribute has a lookback period of the last 7 days, this value will be calculated based on the values of the last 7 days, and then refreshed on a daily basis.
The lookback period for computed attributes is a rolling lookback period. For example, if a first time evaluation occurs on October 15 at 12AM UTC, the a lookback period of two weeks would retrieve all the events from October 1st to October 15th, refresh in one week’s time on October 22nd, then retrieve all the events from October 8th to October 22nd.
Fast refresh
Fast refresh allows you to keep your attributes up-to-date. Enabling this option lets you refresh your computed attributes on a daily basis, even for longer lookback periods, allowing you to rapidly react to user activities.
Next steps
To learn more about creating and managing computed attributes, please read the computed attributes API guide or the computed attributes UI guide.