I can offer you an example. The major database of abstracts in musicology is called RILM – Répertoire International de Littérature Musicale. Nowadays, you get it on CD quarterly, and you do a keyword search.
Previously, though, RILM was distributed as two big quarto volumes. One of them was the keyword index—you look up the keyword, and get a list of six-digit abstract numbers. The other was the article abstracts—you look up the six-digit number, and get an abstract and a citation.
So doing research with these was tedious. You’d start with the most recent volumes that your library had, which was usually 3 or 4 years old. You’d look up your keywords, and look up abstracts, and hope you didn’t follow too many false leads. You’d be lucky if looking up a dozen abstracts for a keyword led to a single useful abstract. And then you’d repeat the whole thing for the prior year.
I tried doing this once when the computer system in my graduate school library was down. I finally concluded that even if it took two days for the computer to be back up, it would be a better use of my time to just wait.
Also, this is why graduate school programs put a lot of emphasis on being current with recent research—nowadays, when the CD database is updated quarterly, you can find some things almost immediately. Previously, when RILM (for instance) took three or four years to come out, the only way you’d know about more recent work was if you had read the journal and remembered seeing it.