XPath is a notation for addressing information within a document. That information could be:
An "executive summary" of a longer document.
A glossary of terms whose definitions are scattered throughout a manual.
The specific sequence of steps, buried in a large reference work, needed to solve a particular problem.
The customized subset of information that a particular customer subscribes to.
All the sections and subsections of a book that were written by a particular author or revised since a specific date.
For documents holding information from relational databases, all the typical queries made of relational databases: a particular patient's medical records, the address of the customer with the most orders, the inventory items with low stock levels, and so on.
For documents that are containers for document collections, all the typical queries made in a library catalog or on a website: articles about Abyssinian cats, essays on the proper study of mankind, etc.
A programmer working with an XML-aware programming or scripting language could write code to search the document for the information that meets the specified criteria. The purpose of XPath is to automate this searching so that a non-programming user can address the information just by writing an expression that contains the criteria.