Summary

Today, developers using $search facets building multi-select filtering experiences have to run N queries to calculate the correct facet counts, where N = number of filters. This can result in customers running 20+ queries, which incurs additional pressure and cost to the system and is unacceptable for customers running at scale

Motivation

How likely is it that this problem or use case will occur?

Any search application with a multi-select filter experience will face this problem; estimated about 20% of all Atlas Search users.

If the problem does occur, what are the consequences and how severe are they?

Workaround is to run N queries to separately calculate facet counts for each query

Acceptance Criteria

What specific requirements must be met to consider the design phase complete?

Driver builders should support the new option `doesNotAffect` in the following scopes:
 * equals, in, range, and compound operators in ALL compound clauses and at ANY nesting level for now

See Syntax.