Classification Sets Performance When Using Multiple Report Suites

Hello everyone,

With the upcoming migration from traditional classifications to Classification Sets in Adobe Analytics, I had a question regarding performance optimization.

Previously, when using classifications, we had to create them separately for each report suite, even if the same variable existed across multiple report suites.

With Classification Sets, we can now create a single classification set and apply it to multiple report suites, which simplifies governance and management.

However, the official adobe documentation mentions two important points :

• Processing time improved from ~72 hours to ~24 hours
• Performance mainly depends on the number of unique key values, especially when combining variables from multiple report suites into a single classification set.

Scenario

Suppose we have the same variable (for example an eVar) across 8 report suites.

Instead of creating one classification set covering all 8 report suites, would it be better from a performance perspective to create:

• Two classification sets

• Each covering 4 report suites

The idea is that the number of unique key values processed in each classification set would be smaller, which might improve rule processing performance.

My Questions

1. Does the unique key count get aggregated across all report suites within a classification set, or does Adobe internally deduplicate keys?

2. From a performance standpoint, is it recommended to split high-cardinality variables across multiple classification sets instead of grouping many report suites into one?

3. Are there any best practices or thresholds for unique key counts when designing classification sets?

Would really appreciate insights from anyone who has already tested this in production.

Thanks!

Reference:
https://experienceleague.adobe.com/en/docs/analytics/components/classifications/sets/overview