Amazon Glow

If you’ve ever tried to hold the attention of a child over a video call, you’ll know how frustrating it is for both of you–conversation stalls quickly because no matter how high quality your relationship is in person, there’s only so much to talk about. Glow’s magic is that we give child and adult activities to do together that’s fun and engaging for both.

TL;DR I joined a team with a visionary hardware project and noble mission, and helped build a solid design strategy to get us from good ideas and intentions to a viable, lovable product.

A project in churn

Glow was approved for production in 2016, 2 years before I joined in January of 2018. Its mission has remained consistent to this day. However, the project had been churning on the design side with many false starts on how to tackle its core mission. With two years under its belt, the product as a whole wasn’t much closer to launch than when it started.

At the time, three “Activity Categories” or “ACs”composed the heart of the product: Reading, Art, and Movement. Outside of the system OS and Launcher, these ACs are what users spend their time doing on the device, similar to apps on a smartphone. The design team was small and fairly junior, consisting of a visual designer, 2 UX/General designers, a design technologist/prototyper, and a researcher. 

The team had been asked to deliver designs for all three, working on one small feature at a time, and hoping it’d add up to a great whole. This can be a valid way to build a product, and in fact is a good way to build productivity software. However, for a consumer device, establishing a North Star experience so you can progressively build to that was critical. Worse, the project leadership would frequently change its mind about priorities and inject feedback that would make the team take sharp and decisive–but under-informed–turns, negating previous work.

Narrow and Deepen

I was extremely lucky that the manager I worked for in the wwOps HCD Studio also joined the team at the same time. He’s a phenomenal design leader and we had years of shared trust to stand on. Together, we quickly formulated a strategy. We would put the brakes on the entire project and dive deep on a single AC with the entire team. The goals of this plan were:

• Teach the team “how to fish”. Each of the members had decent experience and were all very talented. However, they hadn’t had a design leader yet in Glow and hadn’t been walked through an effective design process. This process let people learn by doing, all while retaining a sense of ownership.

• Deepen thinking. The team was continually getting blown around by new ideas and directives from leadership. While these thoughts were all good, because they didn’t have a deep foundational understanding of any one AC–especially as it related to their users–the team was over-responsive and became randomized. 

• Make meaningful progress. Because of churn and randomization, each Activity Category had restarted multiple times, frustrating leadership and individual contributors alike. By reducing breadth and increasing depth, we were able to start at the top of the funnel, combine ideas, make prototypes, validate, and ultimately complete an entire Activity Category in a matter of months.

• Novel research for a novel product. Glow addresses two groups of users that are well outside of the mainstream (elderly family and social experiences for very young children), and what we’re attempting doesn’t really have a precedent or direct competitor. The team had relied on a panel of experts with academic expertise on these groups, but had not done any research themselves. The layer of abstraction between academic knowledge and product application caused confusion, misinterpretation, and vast differences of opinion. To exit this stage, we did 3 rounds of user testing, concluding with a highly successful test with external users.

• End to End. It seems hard to believe, but prior to this point, no one had actually envisioned what it would look like to walk through an entire session from the moment the user started a session to when they put it down. 

• All of us. Despite working with each other for months, the team hadn’t worked together. The UX designers were expected to own the ACs with specialists consulting horizontally. While this is a good end-state model (and the one we have today), because of the massive parallelization of work, no single segment of the product had a cohesive team. 

My Role

I was the Principal User Experience Designer leading the teams through each phase of the plan: plan, brainstorm, concept, rapid prototype, validate, down-select, full prototype, validate. I walked a fine line between driving, facilitating, and participating. It was important that the team didn’t feel like I was overriding them; if every member of the team didn’t feel empowered, the exercise would be a failure. I was able to create a positive environment where everyone felt they could contribute and be heard. 

Our first big win

We exited the design swarm with a “path to green” on AC1, Reading–a single synthesis of our ideas we can go build a prototype of and test. Our final test, which was the product’s first with external users, was a smash hit. We exited with all participants rating the experience well above our CSAT threshold, but more importantly, it was the first time customers validated the importance of the product’s mission. It actually made one of our participants cry happy tears!

Killing a zombie

With our first viable product defined and the team now calibrated to what “good” looks like for design process, we could broaden the scope of work again. We put our second-most experienced designer in charge of the Art AC. The rest of the team carried Reading to its conclusion and I took a look at the previously-started Movement AC. 

AC3, Movement was difficult because while our hardware could detect motion and project onto the floor, they couldn’t create an activity that was satisfying to users. None of them saw the same level of engagement as intimately sharing the Playspace. The time and effort it would take for an uncertain return was more than we could afford–the recommendation to move on was the only right one. This was a somewhat difficult viewpoint to sell because Movement is actually a significant part of why the project was greenlit in the first place. However, we did make the hard call to cancel AC3 and reinvest those resources into more promising areas.

Kicking snowballs down hills

My most valuable skill is to take an ambiguous space, dive deep into what users will find satisfying, and develop a multi-phase strategy for developing an MVP (or, more accurately, MLP–Minimum Lovable Product), v2, and v3. This is exactly what I did for Board Games and Video Games. I basically repeated the process established in AC1 Reading with fewer people involved after initial concepting. For each, I worked with one other designer, then added new hires as we went. I delivered the first viable products for each new space, exiting with similarly high test scores.

As before, with the Activity Category in good shape, I was able to hand off the reigns to the more junior designer and focus on the next undefined space.

Key Deliverables

Animatics and prototypes

While mockups are a critical to really understanding how a design is working, Glow’s unusual screen arrangement made it even moreso. It became really clear over the course of our work that lightly interactive prototypes or animatics that can feign interaction would be a key to testing and communicating our ideas.

Human Interface Guidelines

Over the last two years, the team has expanded dramatically as we ready our final push to production. We’ve brought on new staff to own and develop each Activity Category I’ve kicked off, we’re hired contractors, and we’ve even given great creative control to discrete chunks of Activity Categories to external developers. 

To coordinate this quickly-expanding constellation of contributors, I created the Amazon Glow Human Interface Guidelines. This colossal doc contains everything any designer would need to start designing any aspect of Glow, from OS/Launcher to Activity. I didn’t author every word and image (though it was about 70%), but was responsible for driving, outlining, curating, editing, reviewing, and driving signoff for the whole thing.

Here is a zoomed out view of one of my proudest achievements: our Human Interface Guidelines.

Design Document / Activity Spec

Each Activity Category contains multiple activities, for which each has a Design Document. This document contains wireframes, mockups, animatics, wireframes, and detailed text descriptions.