However, metadata can also be expensive and time-consuming, especially if the human element of entering the information is the primary means of getting the data into the files. Automating the populating of metadata fields is desirable, and with new technologies this is achievable, but care must be taken to ensure that the right information is incorporated into an asset. If a user is searching for a specific thing, the system needs to know that the result it returns is the right one, and often this can only be done through human intervention when adding metadata. To facilitate this, some machine-readable metadata fields, such as technical information about a file, can be automated when ingested into a DAM, but descriptive information can be much more subjective. This conceptual thinking around the description of a file can limit its discovery, as can the lack of knowledge of it in terms of the individual populating the descriptive metadata fields. This is where drawing on the ideas of authors such as Bearman [5] regarding users being able to describe documents because they are the ones with the relevant knowledge, can be very useful, but depending on the flow of work of the institution, this may also prove impractical. Regardless of the metadata schema used, it should be properly structured to ensure that it facilitates discovery and displays all relevant information for both the file and the object.