The factor affecting relevancy score by number of documents is Tf-Idf (term frequency-inverted document frequency). It is calculated to measure importance of every words in a query.
For example, a query is composed by word A and word B. If word A appears in most of documents in a collection but B appears just in small number of documents. Then, documents which contain word B many times are relevant for the query instead of A.
e.g. Doc 1 is A A B A, Doc 2 is B A B B, then Doc 2 is higher than Doc 1.
If the customer needs to disable it, he can set „Influence of the static score to the total document score (0-100):“ to 100 on admin UI of the search collection.
However, if it is disabled, only static score e.g. document date and number of link (web only) affects search relevancy.