Before you can even think about converting anything, you need to get the sources to convert. They are usually obtained after consulting with the publisher or directly from the author of the book in question. Regardless if the source comes from the author or the publisher, it's good to have both addresses for later feedback.
Ways to get the source include traditional media such as floppy disks and tapes, or electronic ways such as email or FTP. In the first case, it's very important that the media is in a format that can be read at the converters site. E. g. if a disk is shipped in Macintosh or Amiga file format, it's highly unlikely anyone can read it. The least common nominator for this job is probably still the MS-DOS filesystem, with files compressed in some way to retain the filenames, their case and the directory structure. The same goes with tapes, where one should agree on block size and backup software used before attempting to ship anything.
Problems can arise when using an electronic transport way, too, though. When sending through FTP or email, the files should also be compressed for the reasons given above. If the network in use is too slow, has a high rate of packet loss or connections are terminated often and on a random basis, the archive files containing the source should be chopped into little pieces of 1 megabyte, 100 or even 10 kilobytes, and each of these little pieces can then be transported seperately.
When getting the sources for the Weich-book, this method had to be applied, as the author of the sources lived in Austria, and the network between austrian schools and the german education network was then routed via the EBone network, going from Regensburg to Vienna via Duesseldorf, Amsterdam and Paris, and it was impossible to transfer the 15 megabytes file via FTP in one piece. Several attempts have been taken during all day- and night times, but no chance. The transfer always terminated after some random time. The file was then cut into tiny pieces and send to Regensburg via email, using the automatic retransmit-on-failure mechanisms employed in the mail protocol (SMTP). With this method, the little packets arrived one by one, slowly, though, but surely, until all pieces were there and could be glued together to a big archive again, which then could be processed. With today's Ten34 network connection european educational networks, this problem wouldn't arise anymore, but still it was a problem at the end of 1996!
After getting the source for a document to convert, it's important to check if it's complete, i.e. containing the files for all chapters, index, in case of TEX documents any additional font, class and style files as well as any bibliographic data and information on the order in which to process all the files, if necessary. Any omissions should be reported back to the supplier of the sources immediately, so these can be received as fast as possible. Also, if the source format is unknown this kind of information should (really! :-) be obtained, together with which tools may be helpful, e.g. with which software images were created, which platform the author has used, any special configuration, etc.
What's left to get is a printed version of the document. This can be received usually either from the local library, or if unavailable, often the publisher will supply an issue. The printed version is important before and after the conversion: before, so one can see what's coming up without going through all the source files, and having to figure out a probably non-trivial building process. During and after the conversion, the paper version of the document is most helpful to compare if the converted document is similar enough to the printed version, and if every detail is realized in a satisfactory way.