Cex is the name of the Correos Express's voice assistant, a project launched in 2020 using voice recognition technologies to fill shipping forms in order to accelerate the form completion process and provide the company a marketplace differentiation.
Problem statement
The company technology department direction wanted to provide a differentiating element by bringing voice recognition technologies to their iOS and Android technologies, the form completion through the native apps was low compared to their website. There were also conditions defined about the way the users could fill the form that made this project a real challenge.
The problems we face at the beginning:
- The form had 7 steps an multiple conditional fields
- The technology and the media was something we never worked on
- The fields can be filled in an aleatory order
- The system should recognize the elements left after the first interaction and ask for them
From the list of options available in the market, the team choose to use Google's Dialogflow, that could be used in both OS. This choice was an additional risk, the technology was in a beta stage at that moment.
Getting the knowledge
After we discuss and decide the technology, there were many questions left:
- The user need to fill two directions fields, how we could propose an easy a reliable way to dictate them?
- The complete phrase in order to fill the 7 steps was extremely long, how we can provide easy steps to complete the form using the voice?
- What is the most direct way to obtain the information?
- What happens if the user need additional information?
The book "Designing Voice User Interfaces: Principles of Conversational Experiences" by Cathy Pearl, was the help I needed in order to understand the type of vocabulary, tone, and cadence needed to request the information and provide a good user experience.
Since one of the project requirements was that the users can provide the information in random order, we also use the card sorting method in order to understand the logical order and grouping of the information in order to cover first the most common patterns.
Additional deliverables
In addition to the new functionality, we were also working on a redesign of some screens of the native apps, so there were plently of elements we created during the project.
- New screens for the voice assistant showing the dialog in real time
- Dedicated page in the settings section of the app for the voice assistant functionality
- Images and mockups for the Google Play and Apple stores
- Informational videos living inside the app and Youtube for marketing purposes
We identify som use cases where the voice assistant voice need to be muted, or the ambience sound was loud and the users need textual cues about how the conversation is going. With this in mind we designed a bubble chat screen, showing the questions and parsing what the user said, in order to provide feedback (also to prevent errors).
You can find the link of the promotional videos we designed here:
Conclusions
What I learned from this project:
- Design for voice interfaces it is not a easy task (at all)
- Deeper knowledge abour conversational design
- Usage of dialogflow and awareness about the technology stack