Have you ever built a feature you thought would be perfect, only to find users completely misunderstanding how to use it? 

This is exactly what happened to us when designing a search feature for an app targeted to elderly users with hand dexterity issues.

The context

We needed to design a search feature for a drug delivery device to facilitate users have quick access to information from its extensive user manual. 

The aim was to offer a digital solution that enables users to find answers efficiently using keywords, rather than manually searching through the lengthy manual's hundreds of pages. 

The intended users of this device are elderly people with chronic conditions and hand dexterity issues.

The initial design

We started with a design that offered users a choice: type their search or use voice-to-text. We made this assumption based on the issues this particular user group faces when typing, due to impaired hand dexterity. 

We wanted to offer them the choice first, as we believed to be empowering them by doing so. 

Our secondary main goal with this design choice was to highlight voice-to-text as a viable alternative, as many users might not be familiar with the dictation feature on their phones. 

So instead we built it into the interface, offering them the opportunity to make a conscious decision to use it as a search input. It made complete sense to us.

The unexpected outcome

However, during usability testing, it completely backfired. Instead of using short keywords, almost all participants used long, conversational sentences as search terms, both when typing and speaking. 

This completely unexpected behaviour resulted in them not finding the content they were looking for at all, leading to frustration and confusion.

The learnings

When analysing the test results, we assumed this was going to be their natural search behaviour due to an unfamiliarity with technology and the way internet search works. 

Their use of sentences like “can you help me” and “I would like to” in front of their long sentences informed this belief. To make matters worse, the search engine being used to process our users’ search terms was failing to process the keywords contained in their long sentences. 

Had it worked, it would not have mattered if participants used long sentences to search, because it would have given them accurate results anyway. So, in a way, our design failed our users due to a combination of technology and interface.

The redesign

Based on these findings, we redesigned the app, ditching the pre-search input method selection and centering the screen around a visible search bar, similar to popular search engines like Google, with a big microphone button to show the option of using voice. 

We hoped that this would encourage them to use concise keywords, but assumed they would still use long search terms anyway, as this was the observed behaviour last time. 

To accommodate for that behaviour, we changed the search engine to one that would definitely identify and process keywords within the context of long sentences. Through extensive internal testing, we verified that this was true.

The result

The redesigned app with a simple search bar was very successful. This time, all participants used short, keyword search terms, successfully finding the content they needed.

This negated our assumption that elderly people would always use long sentences to search. Instead, we learned how much of an impact interface design has on communicating to users their expected behaviour.

The takeaway

This case study shows the profound impact that design choices can have on user behaviour, and highlights the important role of user research in any product design cycle. 

It’s a testament to the importance of understanding our population’s mental models and testing assumptions before investing time and money into development and implementation.

The bottomline? Always challenge our assumptions. Look forward to being wrong, because the faster you find out you are, the faster you will learn from it!