Content is managed and provided to users over a communications link using a differencing engine. The differencing engine is configured to selectively cause content to be refreshed in a cache. Specifically, the differencing engine is configured to detect whether a more recent version of a data item is available, and if so, delete a current version of the data item from the cache and retrieve and store in the cache the more recent version of the data item. Content is selected for refresh based upon a set of one or more selection criteria. The selection criteria may include, for example, the source of content, the size of content, the age of content, the type of content and users to which the content is being provided.