For an organization working mainly with xls, pdf and doc files (but also other MS Office tools), how should it deal with metadata? How can we scan all documents and strip the metadata from them?
Metadata can potentially provide information that an attacker would find valuable. For example, the author value can reveal the organization's username convention, which could be later used in conjunction with password guessing or social engineering. The software used to create the file can also be potentially usable for an attacker if the target is using out of date software with known vulnerabilities. The best practice would be to scrub metadata, but it comes down to a business management decision of whether the risk is significant enough to spend time removing the data. There are several easily available tools for metadata inspection and removal.
Exiftool is excellent at viewing/editing document metadata. It's also multi-platform -http://www.sno.phy.queensu.ca/~phil/exiftool/
You can run it with the recursive switch so it can process all files in the current directory and all subdirectories. You could even write a bat one liner so it will be a set and forget operation to cleanse all metadata in the current folder and all subfolders.