The following information/clarification should be added to $exists documentation, to greatly reduce repetitive questions support is getting about the performance of exists :

add a new section to the $exists page-

  Improving $EXISTS performance

when running a query for retrieve/count documents where

field:{ $exists:true}
even though the field we are querying by is indexed , the all the collection pages will be examined , instead of just scanning index

This is highly inefficient and significantly degrades query performance

For example :

db.collection.countDocuments({auditDate:{$exists:true}})

(query meaning - search for documents where the auditDate field exists in a collection (either with a value or with a null value)

To make such query more efficient - a sparse index should be created over the specific field.

example for creating a sparse index:

> db.getCollection('collection').createIndex( 
    { 'auditDate': 1 },
    {  name: 'auditDate_sparse_index_1', sparse: true  }
)

When the sparse index is in place, such a query will run exclusively using the index with increased efficiency and speed.

an alternative query to using $exists:

(Relevant for use cases where a user is looking only for documents where the field we are querying by is not-null)

using a standard index (no need for sparse index) , run:

 db.collection.countDocuments({auditDate:{$ne:null}})