Anything other than a very simple agreement (e.g. NDA) is probably going to consist of more than one document.
Examples:
The information that you care about is scattered across multiple documents. That information only makes sense in the context of the whole agreement, and you’ll either be missing or working off the wrong information if you look only at an individual document.
This gets even more dangerous where an order of precedence applies over terms that conflict between documents, such as in amendments or riders.
So when it comes to building a contract database, any subsequent steps — especially data capture — aren’t going to be useful unless you first do the work of organising your documents into agreements.