With hundreds of thousands of product offerings across the life science space, users need a quick way to identify relevant products based on a wide breadth and depth of search criteria. While the site search function was the primary method for users to find products, research indicated users were encountering two key problems in the search experience; search terms that returned too few or too many result sets and result sets that made it difficult to disambiguate product attributes.
This project was initiated by business stakeholders based on a user research project in which customer groups were observed searching and shopping for products on our ecommerce channel as well as competitors.
With a catalog approaching 500,000 products from a wide range of product categories and applications, it can be difficult to cater search results that support the diversified needs of users. Chemists, Biologists, Pharmaceutical Researchers, & Manufacturers all have different care-abouts, and use different types of product data for purchasing decisions. Because this product data can be very localized to a specific product category or application, users were having trouble identifying these key product attributes when sifting through search results.
Someone searching for syringe filters has specific attributes they are looking for (diameter, pore size, sterility) while someone searching for antibodies will be concerned with different attributes (application, species, citations).
The current state of search results displayed a concatenated list of attributes for each product which make it difficult to visually scan and parse in a results page containing dozens of very similar products.
Research & Design
User research revealed that customers tended to be overly specific or general in their search terms. Often overwhelmed with the volume and noise associated with a results of an overly broad search term, users would ignore what they perceived as non-relevant filters, and proceed with a new search term that often overshot the mark, and failed to return the expected range of products the user expected.
Filtering results by relevant product attribute types helps, but with a large catalog and somewhat overwhelming list of different attribute types, analytics revealved that filter usage in search result pages was greatly under-utilized.
Since analytics revealed that users rarely applied category filters to their results, one option we explored was to suggest categories in the typeahead search form results with the hope to increase the discoverability of relevant category filters which would return more relavent results based on a familiar category, technique, or application.
Product category suggestions in the search site's search lookahead could provide more visibility into a relevant scoped results for overly broad search terms. This pattern is implemented in other ecommerce sites, but was never considered in-house due to how product data was organized and indexed. We quickly learned that to implement a similar pattern, we would need to plan and orchestrate changes to our product information data as well as code.
Results from a scoped product category can provide result templates (column headers) that expose key attributes needed to make a purchase decision.
Category suggestions offered in our search form lookahead results had the potential to scope product results to a specific type of product or application, but the results would still be presented in a standard template that displayed a product number and a description field. The description field data usually displayed a string of keywords and data a product manager deemed important for a product, however this data could widely among similar products, and often concatenated data points into a long sentence that was difficult to compare to other products when scanning a search results page.
Because different product categories have distinct attributes that are important to users, we worked with product management teams to test a pattern in which search results could be templatized to highlight those key attributes, and allow users to more easily compare and scan those attributes among search results.
An example below displays the key attributes for two very different product types, Antibodies and Syringe filters. When browsing antibody results, we found that users zero-in on applications, species, and citations, whereas syringe filter buyers look at pore sizes, diameters, and sterility. Adding these attributes in a predictable, scannable layout allows users to more quickly identify products that meet their particular needs.
Rather than a wholesale change to all products, we implemented a scalable solution with a few product lines to configure attributes and templates for inclusion into search results. This allowed product managers to opt-in to these changes if they wanted to define and be responsible for populating the data.