I DESIGNED & CONDUCTED A SERIES of in-depth interviews to understand why a search tool was frequently returning no results - with an unexpected and relieving outcome.
Role: researcher, recruiter, project manager
Keywords: user research, interviews, recruiting, search tool, search data, qualitative research, mixed methods, collaboration, enterprise software, B2B, platform
Summary
NEED
Product teams at Procore, a B2B construction project management software, discovered a confusing and high-priority issue: 20% of searches conducted in the search console were not returning any results. I was the sole UX researcher brought in alongside a data team to help understand what was going on.
The main concerns were that the search algorithm may not be producing accurate results, or that there were documents and data within Procore that were difficult for users to locate using search terms.
PROJECT DESIGN
I sat in on multiple Data Team meetings, where they were doing an independent analysis of no-results search terms. This helped me understand what types of search terms were most likely to be causing no-results, and form some hypotheses around potential causes.
I recruited Procore users who had recently experienced the no-results issue in the search tool and received permission from them to pull their recent search history, which I then brought to their interview. Together, in one-on-one sessions, we looked at their recent search terms and I asked them to recall and describe what they’d been looking for at the time they searched for each item.
RESULTS
The results of the interviews were unexpected and fascinating!
While most searches ask “Where is this term?”, many project managers were actually using the search tool to ask, “This term isn’t anywhere in our contracts or documents, right?” Then, when the search correctly returned no results, the project managers were validated in their assumption. In other words, I was able to determine that a large majority of the no-results search events were accurate and indicated that that search tool was functioning in a desirable way.
A smaller portion of searches did indicate that we also needed a couple of quick fixes: improve our construction-specific search dictionary, and improve flexibility around search term formatting.
IMPACT
As a result of this finding, product & engineering teams saved months of work that would have been spent solving a non-existent problem. Product was able to modify the benchmark for acceptable no-results events, and prioritize the quick fixes for the search dictionary & formatting flexibility in the immediate roadmap.