Faceted search FTW
What was the problem?
Our customers upload tens of thousands or more documents to understand the unstructured customer feedback they contain. And just about every document customers upload comes along with a variety of metadata, like "purchase date", "store number", "department" and more. Customers expectations for metadata in our product were simple. They wanted to filter their documents by that metadata.
How did we approach this design?
Our users are product managers, marketing managers, business analysts and HR leaders. The vast majority of our users has never written a line of SQL, so from the very beginning it was clear the right solution was going to be easy-to-use for our many nontechnical users. I considered a variety of levels of comlexity from our document filters but eventually settled on the faceted search model. This model allowed our design for filtering in the UI to focus on what users most wanted from filters; To drill down into more precise subsets of documents a look for differences
Wireframes demonstrating the selection of date ranges
Wireframes demonstrating the selection of categorical data
Advanced mockup of the document filter pane
Advanced mockup of the document filter pane with active selections
What did we learn?
Extensive usability testing showed that all test subjects knew intuitively how to use the new faceted search, and we have received no negative feedback of note about this approach in several years. This feature was successful enough that we're now building new features off of these filters, such as Compare Period-over-Period.