Migrate Search Term Normalization (Winter '21 Release)

Answers Search Terms in Analytics normalizes query strings (for things like capitalization, leading and trailing whitespace, and punctuation) into a list of truly unique search terms to see what users are really searching for. Going forward, Search Term Normalization in Analytics will use the same method that the Answers algorithm uses to understand the Search Term and generate results, which is more robust than the previous Analytics normalization. This consolidated single method will provide consistent normalization throughout the platform.

For example, both Analytics and the Answers algorithm will now use the same logic for determining whether punctuation is removed or replaced with a space. This minor change will create more consistency between the Answers algorithm and Analytics.

Users will see two changes in their Analytics:

  1. Any reports using the dimension / filter “Search Term (Normalized)” will now use the new search term normalization method behind the scenes for both retroactive and new search terms. This is the default dimension / filter we recommend for search terms.
  2. To accommodate backwards compatibility, we will introduce a new dimension / filter which will use the current normalization method called “Legacy Search Term Normalized”. Customers can use this if they would like to export data with the old normalization method or if they’re just interested in seeing the difference between the two.

Note: Search Term Normalization will be migrated for all accounts in the General Availability release and will not be available for Early Access.

For those who do not wish to migrate to the new normalization method in reporting at this time, please use the Legacy Search Term Normalized dimension / filter in Report Builder. Our goal is to eventually deprecate this legacy dimension / filter so we don’t support multiple methods in perpetuity. Therefore, this dimension / filter should be used to transition to the new one rather than for long-term reporting. If you see any discrepancies you think are regressions or you are concerned about, please post in the Community!

To see examples of using normalized search terms in Report Builder, visit the Answers Analytics in Report Builder unit.