Honest but curious cloud servers can make inferences about the stored encrypted documents and the profile of a user once it knows the keywords queried by her and the keywords contained in the documents. We propose two progressively refined privacy-preserving conjunctive symmetric searchable encryption (PCSSE) schemes that allow cloud servers to perform conjunctive keyword searches on encrypted documents with different privacy assurances. Our scheme generates randomized search queries that prevent the server from detecting if the same set of keywords are being searched by different queries. It is also able to hide the number of keywords in a query as well as the number of keywords contained in an encrypted document. Our searchable encryption scheme is efficient and at the same time it is secure against the adaptive chosen keywords attack.