The number of people who buy services and products on-line have drastically accelerated over the last couple of years. Service and product vendors like IBM and Amazon claim that 50% of their customers begin their journey on the respective on-line sites be it for technical support, maintenance services or new product purchase. Enabling customers to get to the service/product they are looking for in the most efficient manner is an important problem to solve for superior customer experience. In the specific industry of Technical support services, product sites get several thousand search queries every day. These queries vary across a range of technical issues, maintenance requests or new purchases for a suite of hardware and software products. The key to responding satisfactorily to these queries is in understanding the products or the services being talked about. Our analysis shows that majority of the time users either do not specify the product or the service name in the search query, or they use a non-standard, colloquial version of the name. Thus, finding the right product/service by either implicitly using context or fuzzily using the search terms is both a crucial problem to solve for on-line service providers as well as for the customers. Faceted search and keyword-based search does not work when there are partial or incomplete queries, and typographical mistakes or absence of product information. In this paper, we build a fuzzy-based taxonomy service that (a) extracts the keywords from a user query, (b) handles partial queries by auto-completing in context, (c) utilizes user browsing context and then (d) uses a fuzzy-based similarity metric to retrieve relevant products/services and their associated information from the product taxonomy file. We show on real user queries that by using the service we are able to get 90% of the product names accurately for a technical service provider, when they are either not mentioned or incomplete.