When I first came to Rollio, there was already a working version of the app available for customers. This had been built by the developers and engineers after being told to base the app on a simple messaging/text interface. No design or testing had been done, so I discovered a multitude of issues. 
After getting past some of the more obvious UI issues, like the tiny font and poor use of white space, I quickly discovered some major UX problems. The app was very difficult to use.
It took about a week to learn how to use the app myself! I found there was a high cognitive load for the user. Rollio didn't always ask clear questions, so it was up to the user to memorize a lot of steps beforehand just to get through a simple task.
I knew I would need to rework Rollio's entire conversational flow and add some features to make tasks easier for the user. Rollio was being marketed as a "virtual assistant," so I needed to give this chatbot some on-the-job training!
 I also needed to determine Rollio's personality and rewrite the chatbot's script.

Competitive + Comparative Analysis
I started by comparing Rollio with other business bots (specifically those that work with CRMs) to evaluate their descriptions, bot personalities, and core users.
Key Findings
The bots who spoke with a "conversational" tone were more pleasant to use.
TrackMaven's bot was too playful and funny. One user who had used the app remarked that this became kind of annoying and was inappropriate for a professional app.

I also did a Comparative Analysis to see if there were any elements from the news or entertainment bots we might want to use for Rollio.
Most conveyed their information in multiple short messages instead of one long message. This made the messages easier to read and usually meant the user would read the entire message.

Chatbot Development
The first order of business was writing Rollio's script. I began with an Excel chart and started building possible conversations around the different workflows. I also included any directions for the developers, such as when to include Action Buttons or when portions of the message would be a clickable link.
I then made user flows to make sure all scenarios were accounted for.
UI Design
I then began designing the main Chat UI. This included Action Buttons, clickable "Smart Messages," and pop-up modal windows. I tried wherever possible to give the user the option to "click" rather than type. These options proved to be very popular during usability testing.
I then worked on the Sign In and "Forgot Password" flows. I also had a special request from the back-end developers. They knew that whenever a user needed to sign in again, the app would need anywhere from 3 to 6 seconds to pull any old conversations from the server. I designed a special modal to keep the user informed.
Next came the Settings menu. I designed icons for the main page. Clicking on these would take you to screens for Push Notifications, FAQ, Support, etc.
I also developed alerts for various system errors, software updates, etc.
Then, I designed a series of Push Notifications. Here are two of them.
Finally, I designed an Onboarding sequence that would be shown to the user the first time they opened the app. It could also be accessed any time from the Setting menu by clicking the "Quick Start" icon.
I was also responsible for other design needs within the company, such as marketing materials, brochures, convention booth design, release notes, and user manuals.

User Manual
Marketing Brochures
Release Notes
Booth Design
Like many other start-ups before it, Rollio shut down due to lack of seed money. We were also having problems with customer engagement.
Many had been so confused by the app's first iteration, they didn't come back for the easier to use version. Plus, the stakeholders felt the amazing technology behind Rollio was the true selling point, and that would get more people on board.
With Rollio, you could send 15+ details about a client at once. Rollio would beautifully parse each of those details and enter them into the correct fields in Salesforce. We had what was called "The Perfect Debrief," which was one long message sent to Rollio that included every detail needed to complete a task. Once the user entered this long message, they could be on their way!
I was finding that users did not want to learn long messages. I had done an all-day session of usability testing and had found that EVERY user wanted to be walked through the tasks by Rollio. They were more interested in being guided step-by-step through the process, instead of memorizing long passages of info and worrying they were forgetting something.
I was in the middle of synthesizing these test results and preparing to go to the stakeholders with my recommendations on how to improve the direction of the app. But, we ran out of time.
I still think Rollio is an amazing product. With the right approach, hopefully Rollio will be helping CRM users again very soon. 

© 2018 Eric Osborn - All rights reserved

Other Projects

Back to Top