Skip to main content

KI-125

Document search missing expected results

Created date

07/11/2024

Updated date

07/11/2024

Status

Backlog

Related work item

PLTFM-2834

Description

The architecture of Salesforce's free text search engine SOSL/Einstein Search is highly susceptible to search crowding. This is because the free text search is done first, without regard to any filters or visibility. The free text results are then truncated to 2,000 records and, finally, any developer-specified filters and user visibility are applied to whittle down the remaining 2000 results. There is no way for a developer to limit the search scope to provide more accurate results. This means that it is necessary to use very unique and precise search terms to get meaningful results.

Visit Salesforce’s documentation on how search crowding works.

We also recommend voting on the idea to improve search results here at IdeaExchange to allow for better on-platform search results.

Affects version(s)

All Medical Information Cloud versions

Impacted capabilities

Inquiry Management

Content Management

Steps to reproduce

  1. Have an org with >2,000 Document Version (mvn__CM_Document_Version__c) records that have the same word or phrase in them somewhere.

  2. Have one document that has the same word/phrase but unique metadata that can be filtered on.

  3. Make a list view that only shows the one document.

  4. Search the list view for the non-unique word/phrase.

Expected Results: The document is found.

Actual Results: The document may or may not be found. The more versions containing the phrase, the less likely it will be found.

Workaround

There are two options for working around this issue:

  1. Avoid free text search if precise results are needed.

  2. Try using more unique search terms that only appear in a couple of document versions when ignoring the filters.

Fix version

N/A

Resolution notes

N/A