Computed Attributes Value Distribution Questions

Question

We cannot get the Computed Attributes Value Distribution to show anything except non-zero values.

This remains true even after confirming that the Computed Attribute is evaluating and complete.

In addition to the Value Distribution showing zeros, each of the shown sample profiles also shows a value of 0 for this Computed Attribute. This initially led us to believe there was something wrong with the underlying logic of the Computed Attribute itself.

However we pulled one of our own ECIDs and checked the associated profile within CDP and we were able to confirm that the profile has a value of '9' for the attribute, which seems plausible

We were recently informed by Ultimate Support that a minimum success rate of 30% is required for Computed Attributes. This is also where we learned that the 30% was not only a minimum threshold for the Value Distribution/Sample Profiles QA piece, but also for getting the Computed Attribute in and of itself to populate at all. This is an extremely high target threshold and unrealistic for many of our real-world applications. We also asked Ultimate Support if and how many clients have adopted Computed Attributes given those requirements and we learned that there have not been many Adobe clients to their knowledge who have, and confirmed that many other clients have expressed similar concerns as we have.

Our main questions are:

1) Why is the value distribution not showing any values, when we have confirmed our existing profile has a value of 9?

2) Why are the sample profiles exclusively showing profiles with a value of 0, when we have confirmed that our existing profile has a value of 9?

3) Is there any way to have Computed Attributes shared with Adobe Analytics or Customer Journey Analytics to enable us to scale sizes for on-site traffic?

AmitVishwakarma · Answer

HI ​@sessoms25The behavior described does not necessarily mean that the Computed Attribute calculation failed. The Value Distribution and Sample Profiles panels are sampling-based diagnostics, not a live lookup of every profile. Adobe documents that:Value Distribution reflects the attribute values available when the sampling job ran.A Sample Profile displays the latest merged-profile value for only a small number of sampled profiles.Therefore, these two sections may not match the value currently visible when you look up a specific ECID. A profile showing 9 in Profile Lookup does not guarantee that the Value Distribution panel has already refreshed to include that value.Please validate the following:Check the Computed Attribute's Last evaluated timestamp and Last evaluation status. Only events available before the last successful evaluation timestamp are included.Allow a buffer of a few hours after the evaluation completes, because the computed value is not immediately refreshed in every profile view.Confirm that the source is a user-created, Profile-enabled Experience Event dataset and that the ECID is correctly linked to the profile. Confirm that the profile value 9 is within the Computed Attribute's configured lookback period and is being viewed under the correct sandbox, field path, and merge-policy context.Compare the exact ECID from Profile Lookup with the ECIDs shown in Sample Profiles. The sample panel may be showing a different sampled population.If the same ECID shows 9 in Profile Lookup but 0 in the Computed Attribute details page after a new sampling/evaluation cycle, this should be treated as a UI/data-consistency issue and raised with Adobe Support. Include the IMS Org, sandbox, Computed Attribute name/ID, exact ECID, expression, source dataset, last-evaluation timestamp, and screenshots.Regarding the reported 30% minimum success rate: the current public Adobe documentation does not describe a universal 30% requirement for a Computed Attribute to calculate or populate. You can ask Support to clarify whether the 30% figure is a tenant-specific guardrail, a sampling/preview eligibility threshold, or a requirement for a particular feature version. It should not be presented as a general product rule without that clarification.Computed Attributes are officially supported for profile lookup, segmentation, activation to Real-Time CDP destinations, and Adobe Journey Optimizer use cases. A direct automatic export of a Computed Attribute from the profile store into Adobe Analytics or Customer Journey Analytics is not described as a standard integration.For on-site audience sizing or personalization, the recommended pattern is to build an audience using the Computed Attribute and activate that audience/profile attribute to the appropriate destination, such as Adobe Target. If the requirement is reporting in Customer Journey Analytics, the derived value should first be made available in an Experience Platform dataset with a joinable person identifier, and that dataset can then be included in a CJA connection. This is a dataset-based design, not a direct Computed Attribute-to-CJA handoff.

Sign up

Login with SSO

Login to the community

Login with SSO