From B & L Blog
Google has spoiled data searchers. Since the search engine is so effective for so many kinds of online ferreting, there's an inclination to believe that we'll always find exactly what we want by typing some keywords in a search field. That's just not so. It's especially not so when it comes to eDiscovery.
The problem with keyword searches in eDiscovery is that they're better suited for retrieving data than knowledge. And knowledge is what attorneys are looking for in eDiscovery.
For example, a typical keyword search returns documents based on "relevance." What's relevance? For the most part, it's how many of your keywords appear in the document. Relevance is blind to how one document is connected to another or how documents may share similar characteristics.
Keyword searches work best on a document set that's stable and relatively static. That's not the case with the ad hoc collection of documents that may be garnered through an eDiscovery order.
The effectiveness of keyword searches on the 'Net is augmented by value-added information—links to other documents, Web site traffic and such. That kind of information doesn't exist for the documents gathered in eDiscovery.
Generally, Web searchers know what he or she is looking for—what's important—so it's easy to sift through a keyword search. That's not always the case with eDiscovery. In many cases, the eDiscovery searcher needs the search results to tell her or him what's important.
Most of all, keyword searches are good at returning what might be the best document in a set, but in eDiscovery, all the important documents must be retrieved. Unless the limitations of keyword searches are recognized, a legal team may spend more time deciding how documents are chosen than which ones are chosen.