Search

Kontor

Overview / Problem

Kontor, a visual search and discovery platform, aggregates professional interior design images from top design firms for designers from around the world to build visual inspiration and mood boards for interior design projects. Kontor not only provides a large repository of images to search through but also has an extensive product catalog so that designers can find the products and their manufacturers in the images they have collected and can source them through dealers directly on the Kontor platform.

Search was one of the largest issues we had to tackle for our users. Users were used to searching through Google Images or Pinterest to find what they were looking for. These results were not always tailored specifically to interiors design so most searches would yield useless image results that were unrelated to what they were searching for.

One problem we had to solve as a company was how to generate relevant search results across interior design inspiration, products, manufacturers and product dealers. This robust search would include searching for images, profiles and product detail pages, all in a single search.

Key Issues

1. How do you make search results relevant?

2. What suggestions can you make to guide users to the thing they’re looking for?

Solutions and Key Takeaways

1. Image tagging

• To support industry specific searches, we had a team of former interior designers and design professionals working to build a robust repository of tags that would fuel relevant search results. Using our content teams extensive knowledge of industry terms and by having these tags attached to every image as metadata we were able to provide relevant search results for our users so that when they entered a search query that included “white” “wood” and “wall” into the search bar they would find results of interiors with white, wood walls.

2. “And” vs. “Or” search results

• I realized that our users were frustrated with certain aspects of Google search and Pinterest because instead of always using “and” (meaning the search must contain all terms) it would use “or” (meaning the search would not necessarily contain every term entered). The use of “or” frustrated our users because it wasn’t returning results with all of the terms that they wanted. I made sure to implement an “and” search so that when our users went to search for specific items it would result in images containing all of the terms entered. If there were no relevant results we would make suggestions by using “and” and show that their “or” search did not return anything.

3. Multifaceted search & autocomplete

• One of the most successful aspects of this features was the multifaceted search with autocomplete. I designed search to be able to handle auto suggest for all aspects of our application including image results, manufacturer and dealer profiles, image collections, and products. Each of these unique aspects were searchable so that if a user typed in “wood” they could see an autocomplete for all relevant items in their suggested search. We found this was extremely helpful for the users since we could not predict what type of thing they were searching for. This enabled our users to quickly search through everything within the application with ease and simplicity.

4. Responsive Design

• A lot of our users were accessing the application on their mobile web browsers so we needed to make sure that the image grid and search dropdown were responsive from desktop all the way down to mobile. It was important to keep the search bar as a prominent element directly under the navigation bar (since it was our most used feature) and not hiding it behind another click so that the users could easily search from their mobile device.