Most indexes are created as a separate document. In contrast, an embedded index is embedded into the document of the book.
To create an embedded index, special codes are embedded directly into the text of the book (hence the name). They can be placed next to the term being indexed, but are more often placed at the start of the paragraph or page. The codes can then compiled into a hyperlinked index. Clicking on the locators in the index will bounce you directly to the page or paragraph containing the information in question.
A number of word processors support embedded indexing, including Microsoft Word, Adobe FrameMaker and InDesign, and LaTeX. I personally will embed indexes in LaTeX and InDesign, (and possibly others if someone asks someday) but not MS Word. Sorry, I just don't like Word and I don't want to work with it. When somebody asks me to work in Word, I refer them to another indexer.
Embedded indexing is most often used to create both print and e-book copies from the same document. It can also be useful if a book will undergo many changes. As long as any edits or additions leave the codes intact, the index can simply be regenerated afterwards, and any pagination changes will be automatically applied. Though of course, if you add a great deal to the book, you will probably want to add new indexing codes to those sections.
The most laborious way to create an embedded index is to insert the codes into the text as you go. The problem with this method is that the Indexer can't appreciate the index as a whole until it's been generated. At that point, the Indexer can see what edits need to be made, but has to return to the original codes to make those edits. The generation and editing process may have to be repeated several times.
This is why Indexers most often use special software to help them embed the codes. That way, the Indexer can happily putter away in their favourite indexing software, creating the index and editing to perfection. Once that stage is complete, the software puts all the codes in the proper places… unless it doesn't.
The reality is that technology doesn't always work the way it's supposed to. When it fails, it can produce tremendous stress on the poor indexer, who more often than not is working under a deadline.
The adventurous souls who do embedded indexing must deal with the added difficulty of working with technology, pay extra for this technology, put in the extra time involved, debug any problems that arise and do it all under a deadline.
For these reasons, I charge about twenty percent more for embedded indexing.