Prototyping with Intention

In our first post of this series we introduced why, and a bit of how, we’re applying experience design to our Open Innovation projects and community collaboration. An integral part of experience design is growing an idea from a concept to a full-fledged product or service. In getting from one to the other, thinking and acting prototypically can make a significant difference in overall quality and sets us up for early, consistent feedback. We are then able to continually identify new questions and test our hypotheses with incremental work. So, what do we actually mean by thinking and acting prototypically?

Common Voice started as a proof of concept prototype and has been collaboratively iterated over the past year

Be Open to Change

At the start of any project our Open Innovation team concepts with the intention that things will change. Whether it be wireframe prototypes or coded experiments, iteration is inevitable. First ideas are often far from perfect… it’s with help from new or returning contributors and collaborating project teams that we’re able to refine initial ideas more readily and efficiently. How? Through feedback loops designed with tools such as Discourse, GitHub, contact forms, on-site surveys and remote testing. Our overall goal being: Release assumptions early and learn from those engaging with the concept. In this way we set our experiences up for incremental, data influenced iteration.

Workshop paper prototypes became coded production prototypes over a 6 week stretch

To continue with our example of Common Voice, we see that this approach was applied in moving from paper prototype to first production prototype. The learnings and feedback from the design sprint exercises helped us realize the need for storytelling and a human interaction experience that would resonate with, well, humans. To achieve this we set out over a 6 week phase to create the experience via wireframes, basic UI design and code implementation. With the help of our community members we were gratefully able to QA the experience as we released it.


Iterate Consistently and Incrementally

With a working prototype out in the wild our team sets their focus on observing and gathering info about performance and usability. In addition to 250+ technical contributors that file issues with feature requests and bug fixes, for Common Voice, our team made time to evaluate the prototype from a usability perspective.

The Common Voice GitHub repository is a hub of collaboration between contributors and Mozilla staff

About three months in we performed a UX assessment reviewing initial prototype designs against what actually made it to production code. Comparing this against feature requests from product stakeholders and contributors, our experience design goal was to understand changes most needed to improve usability and engagement across the site.

This assessment information, combined with usability testing, supported decisions for improvements such as:

  1. Adding keyboard shortcuts to the contribution experience
  2. Improving prompts and progress counters when recording and listening to sentences
  3. Site navigation layout from sidebar to top header
  4. Optimization for responsiveness across viewports
  5. Providing clear calls to action for contribution on the homepage
The next iteration of the MVP prototype based on usability feedback and contributor feature requests

Workshop New Questions

Completing the incremental work allows us to find our way to new questions and needs as a product or service evolves. Along with the feature requests and smaller production needs required of a live prototype, there are larger project strategy queries that can come to light. These are the types of queries you can only learn from experimenting.

Releasing our first dataset for Common Voice was the result of one such experiment. An achievement in itself, it also proved that our concept had merit. The prototype was working! Despite this, in equal measure it also highlighted quality gaps in our data: it could be more spontaneous, such as two humans naturally conversing would allow. It also reaffirmed something we already knew: our data could be far more diverse. Meaning more gender, accent, dialect and overall language diversity. There is an increasing need for a large, publicly open multi-language voice dataset. This has been clear from the start of this project. True to our desire to think and act prototypically we had to choose a single language to focus resources and first prove out the concept. With the successful release of the first dataset we were ready to take on some new questions and keep iterating:

  1. How might we enable a multi-language experience?
  2. How might we increase the quantity and quality of our contributions?

Having already gained integral insights for Common Voice via an experience workshop, we planned another. In January of 2018 we brought together commercial and academic partners to join Mozilla team members, including various expert influencers, to help brainstorm and ideate potential solutions for these questions. The common interest of the attendees? Seeing this unique project succeed. Many had come up against these types of questions in different contexts across their work and were keen to ideate on ways to improve the site.

Multi-language experience wireframes result from a collaborative experience journey and feature prioritization

Workshopping the first question meant determining requirements (what does success look like) and mapping experience journeys to achieve those requirements (see the above image). What resulted was this realization: we have big, multi-feature dreams for the overall Common Voice multi-language experience. To make those dreams a reality we first focused on what was most needed first, providing people a way to contribute in their desired language(s). Other features, like building dedicated language pages and creating a community dashboard, are built into our roadmap. This feature prioritization enabled us to deliver a multi-language experience in May of this year. Reaching this milestone has made the second Common Voice dataset release — which will be our first multi-language dataset release — achievable by the end of 2018.

Workshop session on how we might increase the quantity and quality of voice contributions for Common Voice

In the area of increasing quantity and quality of contributions, the workshop introduced concepts for improving spontaneous speech capture through potential, future experiments. Some examples include enabling spontaneous, conversational style recording sessions on the website; integrations with existing wearables for set session lengths; and a roaming event pop-up with recording booths. This ideation session even lingered in our minds well past the workshop and has prompted thoughts around an opt-in style recording space in collaboration with Hubs, a virtual spaces experiment by Mozilla’s Mixed Reality team.

Relaunched in August 2018 as a portal, the contribution experience is now multi-language enabled

For the current online experience we solidified user journeys that enabled immediate impact of the website and began laying foundation that would enable more robust future experiments. Some of these, such as the new contribution experience and homepage, we’ve already seen land in production as iterations of the Common Voice MVP prototype. Other feature enhancements, like a new profile login experience — which enables contributors to save their progress across multiple languages and view that progress via a new dashboard — have launched this week and are undergoing collaborative QA with our communities. The goal of these features being to improve the human experience while increasing the quality and quantity of voice contributions.

Prototyping continues with the new stat dashboard for Common Voice

With Common Voice we see through incremental, open iteration that our team has been able to intentionally grow from the early prototype. In doing so we are actively working to create more avenues for contribution regardless of language, device or location. Our next post will take a deeper look at how we’re empowering contributions of all sizes, in Common Voice and elsewhere, for Open Innovation.