Because enterprise search is complex, yet badly needed, such companies as Autonomy, Convera, Endeca, FAST, iPhrase, Mercado, Verity, IBM, Oracle, and SAP now offer search tools. Google's new enterprise search appliance is used by National Semiconductor, whose spokesman says it is very good for massive scope. The Google appliance can crawl a huge intranet and still return very good results lists. An analyst says users want efficiency that ensures speed to awareness. This is not difficult to achieve, if companies clearly define the content users will be seeking. However, the process can get bogged down when queries involve a range of values, approximate spellings, synonyms, and multiple repositories and data types. Experts warn companies that implementation of effective search involves challenges, because repositories for structured and unstructured data must be linked. Google-type ease of use in an enterprise search rollout is a reach, and finding search tools that match the user's requirements and skill levels can be task-intensive and time-consuming. Distinguishing between intranet and Internet search requirements is key, and users may need a product that does not depend on keyword matches or document popularity in order to get the desired results. For instance, AMR chose Autonomy software that uses pattern-matching to understand the meaning and significance of information before setting relevancy.