Cozmo Performs
How might we showcase Cozmo’s personality in a way that engages the user?
Cozmo Says, a feature in which users type in words for Cozmo to say, boasted high levels of replayability. For the holiday season, we wanted to ship something new and fun for users while building on top of what made Cozmo Says so successful.
Background
Before Redesign
The original Cozmo Says is straightforward but had surprisingly high replayability.
Tools and methods I used were:
observational studies
app usage analytics
user survey responses
prototyping with the Cozmo Python SDK and Node.js
usability testing
play-testing
I delivered a full-color clickable prototype using Sketch and Flinto to the engineering team.
Tools & Methods
When looking at analytics, our product team discovered that users replay the simple Cozmo Says feature more often than games with levels and rewards or games where the user played against Cozmo.
We hypothesized that children preferred:
playing in an open-ended fashion
being in control
collaborating with Cozmo
and designed a feature with those attributes in mind.
Discovering the Problem
Design Challenge
The major design challenges were:
How might we extend the best aspects of Cozmo Says into a meatier feature that kids replay?
How might we do so with our current team of mostly UX, UI, and app resources, i.e. ship a new feature but recycle existing animations, sounds, or robot behaviors?
We decided that the happy medium may be to give children the building blocks of words and animations but allow them to combine the blocks in an infinite number of ways.
User Testing
Two explorations of the layout of the feature; one focused on sequencing, the other on immediate play
Because kids are very literal and tend to be easily influenced by visuals and colors, we tested ideas in relatively high fidelity
When balancing between immediately allowing kids to trigger animations and planning out a sequence of animations, we found that kids generally hated lists and planning because it reminded them of homework. In fact, it seemed difficult for most kids to think sequentially.
We thus decided to make triggering the animation the main user action, but users could still optionally string together animations.
Users used the text-to-speech feature far less often because now it took far more effort than simply tapping on an animation, so we seeded the final feature with preset sentences.
Task Flow After Making Previewing the Main Action
As part of the validation process, I built a simple “works-like” prototype in Node.js to simulate the functionality of the feature.
Kids filtered animations by their icons instead of words, so we iterated on iconography for clarity.
We wanted to see users’ attention split between Cozmo and the app, which was hard to tease out in visual mockups, so I built a prototype in Node.js that actually triggered robot animations. We learned that kids had a hard time filtering animations by words only and really needed the icons, and that it would be useful to know what animation currently playing.
Kids were far more charmed by human or animal-like animations like “sneeze”, “act like cat”, or “dizzy” over movement-based animations like “move forward.”
Cozmo Performs: Final Design
Design decisions:
The overall design takes inspiration from that of a DJ controller. The DJ controller imagery was chosen to evoke a sense of “coolness” and immediacy, as we learned that children generally do not like planning or putting things into lists and live in the present.
On the left is the currently active animation, and on the right is the entire set of available animations. Above the set of animations is the text box for users to type in things for Cozmo to say. To the right of the text box is the “wildcard” button.
On the bottom is 6 slots, inspired by story spines and Hemingway’s 6-word stories. The idea here is that restricting the number of animations and words kids can string together will prompt them to be a little more thoughtful about which ones they choose.
Feature’s Impact
Average use time of the feature grew from around 300 seconds to 400 seconds.
One of the advantages of shipping a feature like Cozmo Performs is that it is easily expandable; in sprints where we had fewer resources, we simply released a new animation pack to Cozmo Performs.
Cozmo Performs became the de facto feature for users to quickly demonstrate Cozmo’s personality to friends. It also became a favorite tool during marketing demos.