Disambiguating intent is challenging, especially for short, e-commerce queries. Query suggestions and facets can help, but they move the onus of disambiguation to the user, increasing cognitive load. Like a helpful, intuitive sales assistant, Rank Dynamics divines user need with real-time implicit relevance feedback collected by observing behavioral signals in order to build an "at the moment" model of inferred intent. The model is exploited immediately in the current query session to fetch relevant product
Nine years of perfecting the technology has led to impressive results, including +40% improvements in relevance over Google in organic search and +30% increase in sales in e-commerce search. Some of our results have been published by SIGIR and CIKM.
VIEW RESEARCH