Search Assist has two components. First, auto-complete identifies when users need search suggestions by measuring typing speed and responding to hesitations. This feature has dramatically reduced the number of misspelled queries. Second, auto-suggest identifies related concepts, and helps users to move forward (refine), backward (expand), and sideways (related).
Tom explained that this innovation resulted from careful analysis of user behavior and psychology. Their studies showed searchers hadn't become more sophisticated over the past five years, and that users often blamed themselves for poor results. By studying the ways users fail, Tom's team saw an opportunity to help users make the query smarter, one simple step at a time.
This was clearly a good move. Since launching Search Assist, Yahoo! has seen significant improvements in user satisfaction, and a 61% increase in successful task completion. And, in a recent Keynote Benchmark study, Yahoo! took first place in the search assistance category.
Tom noted that thinking creatively about how you define the problem is essential to innovation. There are conspicuous opportunities (e.g., social, multimedia, ubiquitous access), but we must also seek the less obvious possibilities (e.g., making search invisible).
He finds inspiration in such works as Edward O. Wilson's The Diversity of Life. As Tom noted "any system, as it becomes more complex, approximates a biological system." That's an intriguing perspective for someone tasked with inventing the future of search. Good luck Tom!