Helping CloudCall better understand their users
I had the opportunity to work with CloudCall and learn from them as an internship.
Cloudcall are a telephony company that specialises in the recruitment sector. This means they provide SaaS solutions to make it easier for recruiters and sales agents to call their leads.
In my time there i got to work with them on their desktop apps, including excel and chrome extensions, as well as their web portal and mobile app (iOS + Android)
I got the opportunity to work on the early discover phase with the lead designer Goncalo Andrade.
This research was already being conducted when I joined, and it involved a lot of interviews with clients to validate the existing personas, which were based on assumptions at the time.
I started by joining the interviews and taking notes, which I later helped aggregate into findings, and put into a research report.
After that, I got involved with a project regarding an entirely new feature set around transcriptions and sentiment analysis.
In order to remain competitive Cloudcall was looking at moving further into the SaaS space (software as a service). One of the ways to achieve this would be to provide transcripts of calls and voicemails.
This project, also involved discovering ways we could leverage the technology further. In order to achieve this we decided to put together hypothesis and validate them in user interviews, in which we gave the participants a prototype and tasks.
I helped design some of the screens, particularly the supervisor panel. This was a place where supervisors could see the status of their team and watch live transcripts. I also helped write the discussion guide, where we made sure to give the participants a scenario in which they would have to use the features we wanted to test. By watching the go through it, we could gauge the desirability of the features and watch their reaction.
Usability Testing
From speaking with 12 participants, we obtained nearly 900 notes which I then aggregated into findings and a research report.
I was asked to this analysis myself, where I got evidence to validate some hypothesis and discarded some others, as well as learning things we weren't expecting. For instance we learned that,
- Participants didn't fully trust the sentiment analysis done by AI to be accurate, and needed some reassurance there
- Supervisors liked seeing a live transcript of the call an agent is doing, as they wouldn't need to bother or distract them
- Participants need to do their job quickly, so they appreciated some time-saving features we showed the, like:
- Pre-populated notes with suggestions generated by our AI
- Context-sensitive next best action buttons after leaving a call
- Important words and phrases highlighted in the transcripts to facilitate skimming
These findings were refined and organised into a report to be presented to the rest of the company at a town hall meeting. I made sure to illustrate the findings and include highlight clips to help readers empathise with the users. I also linked the raw data so anyone can follow-up in the future.
Next Design Iteration
Once we had this data, we designed the next version where we made changes according to our findings
This was then prioritised with the PM and handed over the the devs, it is currently in the backlog for development but we expect it to be successful and fulfil the needs of the users