Recordly is an app that transcribes audio recordings in real time to assist journalists and other media professionals when they write stories. At initial stakeholder meetings the Recordly team asked us to expand on their user research as well as design a new call recording feature. They wanted to better understand how their users recorded and transcribed interviews as well as make sure their overall concept was appealing to their target audience.
To click through our final prototype, click here, or to see our final annotated wireframes click here.
From an existing competitive analysis we learned that Recordly had one main competitor - Trint - that uses technology to transcribe audio files. Trint is a web-based desktop platform that required users to upload audio files to be transcribed and exported. The Recordly stakeholders told us their plan was to dominate the mobile space and include the recording and transcription in one app.
After getting on the same page with the Recordly team, we needed to learn about our users' work processes and evaluate the existing app. To make the most efficient use of our time we performed user interviews and concept testing together with each user we met with.
We interviewed everyone from Emmy-award winning journalists and TV reporters to entry-level job seekers and graduate students. Our diverse user base helped us gain a better sense of the industry and narrow in on pain points that spanned across many stages of a journalist’s careers.
We learned that:
Many of the people who did not use a transcript said they would if it was less time consuming and tedious. Across the board our interviewees expressed interest in Recordly with one caveat:
In order for a transcription service to be helpful it would have to be 80-95 percent accurate
Early in our research, we had casually tested out Recordly’s transcription software, and our experience made us concerned that Recordly would not be accurate enough. We wanted to get a better picture of how well Recordly would perform, so we compared a transcription from Recordly with a manual transcription of the same audio clip. The results were surprising:
We realized that our suspicions had come true. The Recordly software was only 40-81% percent accurate - far from the 85-90% required by our users. We had to figure out a way to address the accuracy issue while staying true to Recordly’s mission of helping media professionals be more efficient.
Before we could think of solutions to help media professionals, we needed to make sure we all clearly understood the journalists’ writing processes.
The differing persona work styles presented both a problem and an opportunity. It was impossible to choose one as a primary persona as both were common among our user base, so we decided to serve both. We needed to design a product that would help both of our personas capture accurate, trustworthy quotes and information for their stories.
After we defined our problem, we performed brain-writing activities to help us ideate solutions. We came up with three concept designs.
Each concept presented a different feature for Recordly that could add value despite inaccuracies in the transcriptions.
The second idea we tested focused on taking timestamps with the phone or smartwatch during an interview.
When we showed our designs to users, they told us they liked the addition of the smartwatch and extraction tools, but they did not need any help organizing their notes.
We decided to combine the original app with the first and second concepts. The accuracy of the app still would be an issue that the Recordly team would have to address in the future, but the new features would make Recordly effective for journalists even with inaccurate transcripts.
Using our updated design, users could:
With the wireframes in hand we again performed joint usability-concept testing. The tests were heavily focused on usability as may of the features had already been individually tested for desirability.
Overall, users said the combination of features was very appealing. Users:
In addition to our desirability feedback, we came away with several actionable usability insights, including:
We also tested three interactions with the apple watch for taking timestamps, and narrowed our selections down to two that users found least intrusive:
The best part about working with Recordly was the ability to have a major impact on the product. We went in to the project thinking we'd just be making some small tweaks, but ended up designing entire new features based on our research. It was a challenge to present bad news about Recordly's accuracy to our stakeholders, but having research to back us up made our job a lot easier.
If you'd like to learn more, check out my next case study. You'll see how I helped design an app to connect donors and charitable organizations.
